INVESTIGATION OF HYDROGEN CHEMISORPTION ON THE TUNGSTEN (100) SURFACE

BY

STEPHEN PARKER WITHROW

BS Cornell University 1967 MS University of Illinois 1969

THESIS

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Physics

in the Graduate College of the University of Illinois at Urbana-Champaign 1975

Urbana Illinois

INVESTIGATION OF HYDROGEN CHEMISORPTION ON THE TUNGSTEN (100) SURFACE

Stephen Parker Withrow PhD Department of Physics

University of Illinois at Urbana-Champaign 1975

A new apparatus for surface studies has been constructed and placed

in operation The instrument incorporates a monoenergetic beam of electrons

a high resolution high sensitivity electrostatic energy spectrometer

for electrons and ions plus other hardware and electronics necessary to

perform work function measurements Auger electron spectroscopy electron

stimulated desorption of ions low energy electron diffraction and electron

scattering energy loss measurements It has been used to investigate

geometric energetic and kinetic properties associated with the chemisorpshy

tion of hydrogen on a single crystal tungsten (100) surface over a temperashy

ture range from 600 c to -1500 C

Auger spectroscopy data indicate heating thecrystl11 to 2206degK

is sufficient to clean the surface of carbon and oxygen Howeverweakl

structure is observed between 475 eV and 520 eV Possible sources of this

structure have been considered

The work function changes observed at both 60 0 C and -1500 C as a

function of exposure of the crystal surface to hydrogen are equivalent within

experimental error The full monolayer changes in work function with hydrogen

adsorption are 098 eV and 092 eV at the two temperatures respectively

Cooling the clean surface from 2200 C to _1500 C increases the work function

~ 005 eV

Hydrogen ions are desorbed from the dosed tungsten surface following

electron impact The distribution in energy of these ions has a Gaussian

line shape with a peak intensity at an ion energy of 30 eV and a full width

at half maximum of 26 eV The ion yield is found to be coverage dependent

reaching a maximum at work function changes of 018 eV and 011 eV for desorpshy

tion at 600C and -150oC respectively The total desorption cross section

-20 2 0for H+ from the S2 hydrogen state 1S calculated to be 5xlO cm at 60 C

-19 2 0and 12xlO cm at -150 C

A new hydrogen binding state labeled the fast state has been

observed and characterized using ESD techniques This state is desorbable

if the undosed crystal is cooled below room temperature It is populated to

-4 a very low coverage ~10 monolayer by adsorption from the background of

some other t han H T e tota esorpt10n cross cm2spec1es 2 hId sect1on 10- 17 and

energy distribution of the fast state are markedly different than for the S

states

LEED pattern changes observed as a function of hydrogen coverage

are in good agreement with results published in the literature The intenshy

sity of the c(2x2) pattern associated with the S2 hydrogen state maximizes

at a work function change with adsorption of 015 eV LEED elastic intensity

profiles have been obtained for several low index beams These are compared

to theoretical microscopic model calculations and to published experimental

profiles A c(2x2) LEED pattern is obtained if the crystal is cooled below

room temperature Attempts have been made to correlate the appearance of this

pattern to changes in measurements taken in this low temperature range with

other techniques No significant bulk to surface diffusion of hydrogen was

found No evidence was found from work function and electron scattering

experiments to support the hypothesis that the low temperature c(2x2)

pattern results from hydrogen on the surface In addition the fast hydrogen

state is not believed to be responsible for the c(2x2) pattern LEED

intensity profiles from the undosed cold surface c(2x2) pattern and from the

c(2x2) pattern associated with the ~2 hydrogen state are not identical an

indication that the atomic geometry responsible for the two patterns may not

be the same

Electron scattering energy loss spectra have been obtained as a

function of hydrogen coverage for loss energies of 0 eV ~ w lt 40 eV Hydrogen

adsorption has a marked effect on the spectra Fine structure in the data

has been discussed and compared to photoemission data Cooling the clean

crystal below room temperature causes only a slight change in the energy loss

spectrum

iii

ACKNOWLEDGMENTS

I would like to sincerely thank my advisor Professor F M Propst

for guidance given throughout the course of this research and for his knowledge

of surface research which helped shape this work Dr C B Duke for many

helpful conversations and for assistance in the analysis of the data my wife

Lois for her patience and steadfast encouragement my parents and parents-inshy

law for their generous and needed support Paul Luscher and John Wendelken for

their creativity and cooperation in the design of the apparatus and for many

interesting scientific discussions Jim Burkstrand and Chris Foster for their

friendship and technical help on many occasions Bob Bales George Bouck Bob

Fults Bill Beaulin Lyle Bandy and Bill Lawrence for design assistance and

expert workmanship in the construction of the apparatus Nick Vassos and Bill

Thrasher for the imaginative and superior work in materials processing necesshy

sary in the construction of the instrument Gene Gardner and co-workers for

able assistance in electronics Bob MacFarlane and assistants for their high

quality draftmanship Jack Gladin and Jim Gray for an excellent job on the

photography needed for this research Bob Ebeling for his cooperation in the

print shop Chris Fleener for an excellent typing of the manuscript and the

rest of the staff of the Coordinated Science Laboratory whose efforts have

been gratefully appreciated during my tenure as a graduate student

bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

iv

TABLE OF CONTENTS

Chapter Page

1 MOTIVATION AND REVIEW OF LITERATURE bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 1

11 Introduction and Motivationbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 1

12 Review of Literature bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 4o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

121 Introductionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 4 122 Thermal Desorptionbull o 4 123 LEED and HEEDbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 7 124 Electron Scattering Field Emission and

Photoemissionbullbullbull 100 bullbullbullbullbull 0 bullbullbullbull

125 Electron Stimulated Desorptionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 12 126 Work Functionbullbullbullbullbullbullbullbullbull 13o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

1 27 Chemical Probes bullbullbullbullbullbullbull bullbullbullbull bullbullbullbull 14I) O

128 Chemisorption Models bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 15 1 29 Theoretical Treatment bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 19 1 210 Literature Surnmaryo 20

13 Review of Experimental Techniques bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 21

2 EXPERIMENTAL APPARATUS bullbullbullbullbullbullbullbullbullbullbullbull ~ bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 22

21 CGeneral Description bull 22

22 Vacuum System and Magnetic Shieldingbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 22

23 Electron Gun and Gun Electronics bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 25

24 Energy Spectrometer 29

25 Spectrometer Electronics bullbull 45o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

26 LEED Display System 490 bullbullbull 0 bullbullbullbullbullbullbull 0 bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull I)

27 Crystal Manipulator Assembly bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 51

28 Crystal Preparation and Cleaning Procedures bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 57

29 Gas Handling System o bullbullbullbullbullbullbullbullbullbullbullbullbull o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 62

210 Mass Spectrometer bullbullbullbull o 64

3 EXPERIMENTAL PROCEDURE AND RESULTS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 65

31 Introductionbullbullbullbullbullbullbullbullbullbullbullbull 65

32 Experimental Procedure o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull o bullbullbullbullbullbullbull 65 321 Work Function and Coverage Measurements bullbullbullbullbullbullbullbullbullbullbullbullbull 65 322 Primary Electron Beam Energy and Current

Measurements 6811 bullbullbullbullbullbullbullbullbullbullbullbullbullbullbull bullbullbullbull

323 Target Temperature Measurement bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 69 324 Electron Scattering and Auger Measurements bullbullbullbullbullbullbullbullbullbull 69 325 Ion Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 72o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

v

33 Ion Species Identificationbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 74

34 Work Function Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 77

35 u+ Ion Desorption Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 83 351 ut Ion Energy Distributionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 83 352 Desorption Curve Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 93 353 Temperature Dependence of the Fast State a+ Ion

Signal 4) 96

354 Additional Measurements on the Fast State bullbullbullbullbullbullbullbullbullbull 98

36 Low Energy Electron Diffraction Data bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull lll

37 Electron Scattering Data bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 119

4 DISCUSSION OF RESULTS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 126

41 Auger Electron Spectroscopy Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 126

42 Work Function Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 127

43 Electron Stimulated Desorption Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 129 431 Room Temperature Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 130 432 Low Temperature Fast State Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 135 433 Low Temperature Dosed Surface Results bullbullbullbullbullbullbullbullbullbullbullbullbull142

44 Low Energy Electron Diffraction Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 145

45 Electron Scattering Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 161

5 SUMMARY AND CONCLUSIONS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull165

REFERENCES 171

APPENDIX A 0 176 poundI

APPENDIX B 9 c 179

APPENDIX C 0 182 0 bull

VITA 190lilt bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull bullbull

1

CHAPTER 1

MOTIVATION AND REVIEW OF LITERATURE

11 Introduction and Motivation

The experimental investigation of chemisorption of gases onto metallic

surfaces has uncovered in the past decade a baffling complexity of adsorption

geometries and binding state kinetics and energetics A growing interest in

surface phenomena generated by the questions raised in these researches and

by the long anticipated use of the answers has led to the rapid development

of new surface probing techniques and to an increase in the availability of new

technologies compatible with surface research Aided by the new technologies

and techniques chemisorption research has been undertaken on many different

systems (system =solid substrate + gas adsorbate) under a wide range of

environmental conditions Surface theoreticians have proposed microscopic and

phenomenological models with which data have been analyzed and compared A

large experimental and theoretical literature in the field has resulted

The study of adsorption on metal surfaces has among its primary

objectives a detailed description of the physical geometry of the atoms in the

surface layers and a quantitative understanding of electronic and chemical

surface phenomena While considerable progress has been made in documenting

gas-surface phenomena it has been only in the past three years that any

credible claims to understanding specific surface overlayer geometries have

1been made At the same time many basic questions involving system kinetics

and energetics have not been adequately resolved 2

It has become apparent from the failure of the research effort to

satisfy the above objectives and from seemingly conflicting data that adequate

2

characterization is basic to all surface experimentation The need for such

characterization has been strongly emphasized by disagreement between quasishy

similar research performed at different locations Small impurity concentrashy

tions small changes in crystal orientation minor temperature differences etc

often have considerable effects on experimental measurements

A corollary to the necessity of being able to characterize the

surface is the importance of being able to control the surface characterization

In order to compare data it is necessary to reproduce all experimental paramshy

eters accurately The theory to which experimental data are compared is

constrained by complexity to be based on ideal clean substrates with a

definable gas-surface interaction It is desirable to be able to choose the

adsorbate-substrate system and experimental parameters that can also be used

with theoretical models Of course compromises are inevitable

One approach in surface research has been to incorporate several

experimental techniques into one apparatus At the present time this approach

is increasingly more common A combination of surface sensitive probes offers

many advantages It provides for more checks of the data against results

published in the literature It allows for better characterization of the

experimental parameters It allows surface phenomena to be probed in more than

one way in order to eliminate ambiguities andor to reinforce conclusions It

eliminates the uncertainties that arise from data taken at different locations

using the techniques separately hence it allows better correlations between

data taken using the different techniques

In this study several aspects of the chemisorption interaction of

hydrogen with the low index (100) plane of single crystal tungsten denoted

3

W(100) have been investigated The choice of the tungsten-hydrogen system is

motivated by several factors The literature raises questions on the unusual

state behavior observed when hydrogen adsorbs on W(100) questions which our

apparatus is especially well-suited to study In addition the interaction of

hydrogen with close packed planes of single crystal metals including W(100)

is of particular interest to theoreticians since hydrogen is the simplest

molecular gas as detailed in Sections 128 and 129 below recent theories

of chemisorption have been applied to the hydrogen-tungsten system Tungsten

was chosen as the substrate metal because of its relative ease of preparation

and cleaning and because the large body of literature on the hydrogen-tungsten

system indicates wide interest in this interaction

In particular this research is concentrated in three areas Tung~

sten when cleaned and subsequently cooled to below room temperature exhibits

an anomalous centered (2x2) denoted C(2x2) low energy electron diffraction

pattern The possibility that this is caused by hydrogen diffusing from the

bulk or adsorbing from the background has been investigated Secondly the

room temperature and low temperature hydrogen adsorption binding states have

been studied using electron stimulated desorption and electron spectroscopy

Finally low energy electron diffraction has been used in an investigation of

the atomic geometry of the clean tungsten surface and the geometry of one of

the hydrogen binding states

The remaining sections of this chapter summarize literature relating

to this research Chapter 2 details the construction and performance of the

experimental apparatus A presentation of the experimental data is made in

Chapter 3 A discussion of the results is made in Chapter 4 A summaryof

4

the research and conclusions from it are given in Chapter 5

12 Review of Literature

121 Introduction

Experimental investigation of the adsorption of hydrogen on tungsten

has been carried out using field emitter tips macroscopic polycrystalline

surfaces and all the low index orientation single crystal faces including

W(lOO) From the wide range of results it is clear that the substrate exerts

a strong influence on the adsorption properties The question has been raised

whether adsorption models postulated from polycrystalline substrates can be

3extended even qualitatively to single crystal systems Likewise there are

such sharp differences in the chemisorption behavior of hydrogen onto the

different single crystal tungsten surfaces that the specific physical process

governing adsorption on each face appears to be unique to that face This

review will therefore be concerned for the most part with literature dealing

with hydrogen adsorption on W(lOO) the gas-substrate system used in this

research

The following data review is presented in subsections each concenshy

trating on a specific experimental technique or class of experiments Subsecshy

tion 127 includes some gas mixing literature which was considered relevant

and Subsection 129 gives a bibliography of some related chemisorption

theoretical papers

122 Thermal Desorption

One technique widely employed in investigation of the kinetics of gas

456adsorption has been thermal or flash desorption By monitoring the system

5

pressure changes of a species while rapidly heating the crystal inferences

can be made on the existence of binding states adsorption energies and reaction

7 8 order Heating is accomplished by a focussed light beam electron bombardshy

7 9ment or ohmic losses bull

The strongest indication that hydrogen exhibits multiple state

adsorption behavior on W(lOO) comes from flash desorption data Tamm and

8Schmidt 7 and Madey and Yates found for room temperature adsorption that two

large peaks are observed in the plot of pressure of H2 versus temperature

These peaks are centered at 4500 K and approximately 550oK Only hydrogen

molecules are observed to be desorbed from the surface Using desorption rate

456theory the lower temperature (~l) state is found to obey first order

kinetics the peak temperature being independent of initial coverage The

higher temperature (~2) state obeys second order kinetics the peak temperature

moving down from approximately 5800 K to 5500 K as the initial coverage is

increased

In addition to the two main peaks several small peaks comprising a

total of less than 10 percent of the desorbed hydrogen are observed during a

flash when the crystal is cooled to nitrogen temperature before hydrogen

exposure These states are believed to result from adsorption onto edges and

other non-(lOO) crystal orientations 8

The dependence of the desorption energy Ed on coverage is weak

~l 8 ~2 8with Ed = 252 kcallmole and Ed = 323 kcalmole As would be predicted

from this result the ~2 state is largely populated before the ~l state begins

to fill The ratio of the populations in the states at full coverage n~1n~2

8 lOis 2111 Jelend and Menzel point out that the cross section for electron

6

desorption (see Subsection 125) from the ~1 state is considerably smaller

than that of the ~2 state a result not predicted from the experimental flash

desorption binding energies

11Tibbetts using a stepped temperature flash desorption technique

has shown that hydrogen is bound to the surface of po1ycrysta11ine tungsten

W(110) and W(lll) with a distribution of binding energies ranging from

162 kca1mo1e (07 eV) to 415 kca1mo1e (lS eV) Using this technique

desorption at a constant temperature from a distribution of states can be easily

distinguished from desorption from a state with a unique binding energy While

he did not extend his work to W(100) these results suggest that hydrogen does

not bind to the surface at a few energies but rather with a range of binding

energies Tibbetts also shows that the determination of the order of the

desorption process is difficult for adsorption into a range of binding energies

In experiments using a combination of hydrogen and deuterium as room

temperature adsorbates the elemental composition of the desorbed ~1 and ~2

states shows complete isotopic mixing between the states7S independent of the

adsorption sequence or ratios However for low temperature adsorption

complete state conversion is not observed for all the additional small peaks

Where mixing is observed atomic adsorption is suggested Where mixing is not

observed molecular adsorption is inferred

Some theoretical models indicate that multiple peak flash desorption

spectra may be due to binding energy changes caused by adsorbate-adsorbate

interactions12 or order-disorder transactions during f1ash3 13 rather than

actual distinct multiple binding states under the adsorption conditions

7

123 LEED and HEED

Low energy electron diffraction or LEED refers to the coherent

scattering of a monoenergetic focused electron beam from the surface of a

crystalline solid Electrons in the energy range 0 lt E lt 300 eV interact

strongly with the ion cores and valence electrons of the solid The inelastic

mean free path is on the order of 10 X As a consequence electrons which are

observed elastically have been backscattered within the first several surface

layers in the solid making LEED a surface sensitive probe 14 15 The periodic

translational symmetry of the surface leads to the momentum conservation laws

I

= (11)~I ~II + g

= (2mEh2)t sin reg (12)~II

where ~II is the momentum parallel to the surface of the emerging electron and

~ is the parallel momentum of the incident electron The vector g is a

reciprocal lattice vector of the solid surface and reg is the polar angle of the

incident beam As a result of the conservation laws the scattered electrons

emerge in a series of beams called a spot pattern which is a manifestation

of the surface symmetry The beams are labelled by their corresponding g

vectors Information on the actual configuration of the surface ion cores is

extracted from the plot of elastically scattered electron intensity versus

16 17incident energy called elastic intensity profiles or I-V curves

Changes in the LEED spot pattern resulting from hydrogen adsorption

18onto W(lOO) at room temperature have been reported by Estrup and Anderson and

19by Yonehara and Schmidt As hydrogen is adsorbed at room temperature the

8

(lxl) pattern characteristic of clean tungsten changes into a weak C(2x2)

pattern at a hydrogen coverage e of less than 005 monolayer (where e = 1

15 2monolayer at a coverage of 2xlO hydrogen atomsem which is 2 hydrogen atoms

for each surface tungsten atom) The (1212) spots become sharp and intense

with increased hydrogen adsorption reaching a maxima in intensity at approxishy

19 18mately 015 monolayer or e = 025 monolayer (Coverage in both cases was

determined by flash desorption) The (1212) spots split into four new spots

which move apart continuously and upon further hydrogen adsorption eventually

19form 13-order streaks at a coverage of ~ 035 monolayer These in turn

change at higher coverage into poorly developed 14-order features with the

(012) spots missing Finally near e = 10 all extra diffraction features

are gone and the LEED pattern returns to a (lxl) symmetry This series of LEED

pattern changes is reversible with coverage

For hydrogen adsorption at 1950 K the C(2x2) pattern develops

maximizes in intensity at e ~ 18 and splits into four beams similar to room

temperature adsorption However the higher coverage LEED structures third

19 0order streaks and fourth order spots are not observed At 78 K adsorption

of hydrogen produces half-order C(2x2) spots that maximize in intensity at

e ~ 027 Additional hydrogen adsorption beyond this causes the half-order

19spot intensity to decrease but with no splitting of the spots

If clean W(lOO) is allowed to cool below room temperature a sharp

19and intense C(2x2) pattern is observed even though total coverage due to the

background of H2 and CO is less than e = 001 as determined by flash desorption

measurements The intensity of these half-order beams can be altered by

additional hydrogen adsorption or by various hydrogen and oxygen heat treatmenm

9

l9Yonehara and Schmidt have postulated that the extra spots may result from

hydrogen endothermally soluble in tungsten diffusing to the surface Possible

alternative explanations include the reconstruction of the clean tungsten

surface or impurity over layer adsorption from the background It should be

20noted that Plummer and Bell saw no effects upon cooling their field emitter

tip to 78 0 K that they could attribute to hydrogen diffusion to the surface

Elastic intensity profiles for several beams have been measured by

18 21Estrup and Anderson and by Papageorgopoulos and Chen at room temperature

Estrup and Anderson reported profiles for (00) and (1212) tEED lJeams out

to 300 eV The intensity variation for the (OO) beam was found to be indepenshy

dent of hydrogen coverage a manifestation of the relatively weak scattering

cross section of a hydrogen ion core relative to that of tungsten The

additional beams observed on adsorption would be expected to be more sensitive

13to the adsorbed overlayer Estrup found the intensity of the (1212) beams

for H W(100) and D W(100) to decrease as a function of temperature between2 2

300o

K and 6000 K at a constant incident energy but neither obeyed the Debyeshy

22Waller relation Estrup interpreted this result to imply the adsorbed

species are directly involved in the diffraction process

19Yonehara examined the tEED spot intensities as a function of

temperature down to liquid nitrogen temperatures His results are examined

in the data discussion in Chapter 4

The hydrogen-WlOO) system has also been studied using high energy

electron diffraction or HEED23 This technique utilizes the strong forward

scattering of high energy electrons in probing the surface by having the

incident beam strike the solid at grazing incidence The resulting diffraction

10

patterns are sensitive to the translational symmetry of the surface As in

LEED a diffraction pattern with C(2x2) symmetry is observed during initial

hydrogen adsorption With further hydrogen exposure the beams broaden some

splitting is observed and eventually the half-order intensity decreases

completely The width of the observed half-order beams has been interpreted to

indicate hydrogen initially adsorbs into a large number of small square or

rectangular domains each consisting of 20 to 25 atoms with sides parallel to

the [100J directions

124 Electron Scattering Field Emission and Photoemission

The electronic structure of the surface layers of solids can be

characterized by electron emission spectroscopies Such techniques require a

surface sensitive perturbation on the gas-solid system leading to an electron

yield A spectrometer is used to energy analyze the emitted or scattered

24 25electrons In electron scatter~ng a monoenergetic beam of electrons is

directed onto a crystal The inelastically scattered signal containing

quantitative information about electronic excitations at the surface is

measured as a function of energy loss Field emission is a technique in which

application of a high positive electric field to the surface lowers the potenshy

26tia1 barrier for electrons to tunnel out of the system The total energy

distribution of electrons emitted as a result of the applied field contains

bull 27 28spectroscopic data on the adsorbed spec~es Structure induced by the

adsorbate in the elastic tunneling spectrum is thought to be a measure of the

d f i h f h d b 2930e1ectron~c ens~ty 0 states n t e v~c~n~ty 0 tea sor ate The

inelastic tunneling spectrum reveals discrete excitation energies in the

11

gas-surface system In photoemission or ftotoelectron spectroscopy an incident

monochromatic light beam incident on the solid induces emission of electrons by

interband transitions with or without an inelastic scattering after the initial

31 exc i tat~on

The vibrational spectrum of hydrogen adsorbed on the tungsten (100)

32surface has been observed by Propst and Piper using inelastic low energy

20electron scattering and by Plummer and Bell using field emission Propst

and Piper found two peaks in the energy loss distribution E - E pr~mary scattered

at 135 meV and 69 meV for all coverages used up to saturation These results

are interpreted as indicating hydrogen atoms are multiply bonded to tungsten

33atoms No losses near 550 meV the vibrational energy of molecular hydrogen

were observed Plummer and Bell measured the total energy distribution of

field emitted electrons from tungsten (100) as a function of coverage of

hydrogen and deuterium at several temperatures Discrete losses at 100 meV

and 50 meV for deuterium which correspond to the 135 meV and 69 meV vibrational

losses observed by Propst and Piper were evident for deuterium in the a state2

No discrete energy losses at 550 meV or 390 meV the molecular vibrational

levels of H2 and D2 respectively were measured Hence there was no observable

molecular adsorption at room temperature

Additional structure in the field emission total energy distribution

is observed as a function of hydrogen adsorption The clean surface spectrum

has two peaks 04 eV and 15 eV down from the Fermi level which have been

d ifi d h h d f d h h f 34 A h d ~ ent e as ~g ens~ty 0 states assoc~ate w~t t e sur ace s y rogen

(or deuterium) coverage increases between 0 and 5xl014 atomscm2

ie while

the ~2 state is filling the surface states disappear and the field emitted

12

energy distribution changes from that of clean tungsten to a spectrum indicative

of resonance tunneling through a level positioned 09 eV below the Fermi energy

There is a shoulder at -11 eV and an energy dependent background With

further adsorption up to saturation this energy distribution characteristic

of the ~2 state shifts to a broad level centered at 26 eV below the Fermi

level This is interpreted to indicate that the ~1 state does not adsorb on

top of the ~2 state

35The photoemission results of Wac1awski and P1ummer and Feuerbacher

36and Fitton are in agreement with those obtained using the field emission

technique As hydrogen adsorbs on the surface emission from the 04 eV

surface state decreases The level 09 eV down from the Fermi energy as well

as a level 5 eV down remain unchanged by adsorption At full coverage a strong

peak at -25 eV is observed

125 Electron Stimulated Desorption

Desorption of energetic ions and neutrals from a gas over layer can

be induced by electron bombardment This process termed electron stimulated

desorption or ESD37383940 is described below in the discussion of the data

in Chapter 4 In brief intensity and energy measurements of the ion yield can

be interpreted in terms of the parameters characterizing the binding energy

curve of the adsorbed state and adsorption states can be classified by cross

section for desorption_ ion energy distributions primary energy threshold for

desorption etc

ESD has been employed to study adsorption kinetics for the hydrogenshy

41 ~ tungsten (100) system Madey obtained a coverage dependence yield of H ions

13

observing an ion signal during the initial stages of hydrogen adsorption that

reached a maximum at a coverage of e = 017 and then decayed to zero yield

23 2after e = 075 the maximum cross section obtained was 18x10- cm at a

primary electron energy of 100 eV Tamm and Schmidt7 reported a maximum cross

section for H+ desorption of 10- 21 cm 2bull Benninghoven Loebach and Treitz42

10and Je1end and Menze1 obtained qualitatively similar (to Madey) coverage

dependencies on tungsten ribbons which were believed to be predominantly of

(100) crystallites the main difference between these results and Madeys is

the observation of a residual signal at high hydrogen coverage

The cross section results indicate that the a state even though it2

is more tightly bound than the a state provides more hydrogen ions upon1

electron impact The cross section for ion desorption from the a1 state is

d41 10-25est1mate to be 1ess t han cm2 bull

+No energy distributions of the desorbed H ions have been reported

The dependence of the desorption cross section on the primary electron energy

10has not been investigated Je1end and Menzel found their results independent

of temperature for temperatures above room temperature No ESD results for

otemperatures less than 300 K have been reported

126 Work Function

The work function is the minimum amount of energy required to remove

43 an electron from a meta1 An adsorbate may alter the potential barrier for

electrons at the surface and hence change the work function Such is the case

for hydrogen on W(100) Hydrogen adsorbs as an electronegative ion on W(100)

thereby increasing the work function

14

44Using a retarding potential current-voltage techn~que Madey and

8Yates reported a linear relation between work function change and hydrogen

coverage over the entire coverage range for adsorption occurring at a

temperature 0 f 2= 300 0 K Coverages were determined by flash desorption

Total work function change at one monolayer was found to be 090 eV which

corresponds to a constant dipole moment per adsorbed atom of 015 Debye

15 2(assuming a total coverage of 15xlO atomsem)

Estrup and Anderson18 using the retarding potential technique and

Plummer and Bel120 using a field emission Fowler-Nordheim plot45 found a

work function change versus coverage relationship which deviated only slightly

8from being linear in near agreement with the results of Madey and Yates

46 47Becker using field emission and Armstrong using a retarding potential

technique found a work function change for the H W(100) system of approximately2

48090 eV when e = 10 Hopkins and Pender obtained an anomalously low value

49of 054 eV while Collins and Blott obtained a high value of 109 eV for the

same coverage

127 Chemical Probes

Hydrogen adsorption has also been studied using a chemical probe

ie by coadsorbing some additional different gas with the hydrogen and

observing the effect this additional species had on results obtained using

techniques discussed in previous sections

50Yates and Madey found using flash desorption that CO coadsorption

onto a hydrogen monolayer displaces both the Sl and S2 binding states with

near unity probability in doing so initially replacing these states with a

15

less tightly bound hydrogen state After high CO exposure no hydrogen remains

on the surface if the coadsorption is done at room temperature At lower

temperatures only about 10 percent of the hydrogen is desorbed Jelend and

Menzel lO 5l found that the ionic and total desorption cross sections for

hydrogen the sticking probability and bond strength of hydrogen are all

affected by preadsorption of CO or oxygen Yates and Madey3 found that N2

slowly replaces preadsorbed hydrogen at temperatures greater than 300oC Below

2730

K the displacement of hydrogen by N2 does not occur

These results all show that small impurity concentrations of CO N

and 0 if intentionally introduced do affect the resulting hydrogen chemishy

sorption properties These experiments underscore the necessity of surface

chemical characterization

128 Chemisorption Models

7 18 19 52 Several models have been proposed to account for the

experimental results from hydrogen chemisorption on W(lOO) Each of these

models embodies some combination of the following three phenomenological

assumptions 1) The tungsten surface is ideal planar and free of contamishy

nants 2) There are available two types of adsorbates namely hydrogen atoms

53and molecules 3) There are three high symmetry logical bonding sites on

a BCC (100) face usually designated by the letters A B and C These

positions are A a bridge site with 2 nearest neighbors and the hydrogen

equidistant from two adjacent surface tungsten atoms B a four fold symmetric

site with 5 nearest neighbors and the hydrogen located on a surface normal that

passes through the center of a surface primitive unit cell and C the hydrogen

16

located directly above a surfac~ atom with one nearest neighbor There are

15 152x10 A sitescm2 and 1x10 B ore sites~m2 on middotW(100)

The proposed models attempt to explain or at least be consistent

with empirical results presented in Sections 121 through 127 A summary

of salient features of the reported research includes a) a linear work function

15change with coverage b) a saturation coverage of between 15 and 20x10 H

2atomscm c) a coverage dependent adsorption layer symmetry including a we11shy

ordered surface at low coverages d) the appearance at low temperatures of a

surface translational symmetry not equivalent to room temperature clean

tungsten e) adsorption into states which produces a two state thermal desorpshy

tion spectrum with relative coverages in the two states of nQ n = 21 f) a ~1 ~2

coverage dependence for the observed electronic structure including the

absence at high coverage of structure observed at low coverages and g) no

observed vibrational levels which correspond to free H2 molecule

Following is a short review of several proposed surface models

18Estrup and Anderson account for their LEED spot pattern results

using a model wherein the hydrogen molecules dissociate upon adsorption and

migrate initially to form islands of hydrogen atoms located at type A sites in

a e(2x2) structure Dissociation is assumed since the e(2x2) pattern maximizes

15 2at a coverage of 25x10 mo1ecu1escm whereas without dissociation it would

maX1m1ze at 5 1015 mo ecu1es cm2bull e site adsorptionx 1 Although type B or type

could also produce the initial e(2x2) LEED pattern these sites are rejected

15 2because of the following 1) There are only 1x10 B or e sitescm This is

not enough sites to accommodate a full monolayer coverage if all binding is

atomic and if it is into equivalent sites as suggested by the linear work

17

function change with coverage 2) While a complete 14-order LEED pattern is

possible with type B or type C sites it is not possible to produce a 14-order

54pattern with (012) spots missing using B or C sites As the adsorption

14 2proceeds past a surface density of 5xlO hydrogen atomscm the adatoms conshy

tinuously rearrange into long period superlattices with antiphase domains

that produce successively the 12-order spot splitting the 13-order streak

(only partly ordered domains) the 14-order pattern and finally the (lxl)

full coverage structure

This model has been criticized for not accounting for the two

observed binding states and for predicting a maximum in intensity of the half

order spots at e = 025 these beams maximize at less than that coverage In

addition the authors themselves questioned the idea of well-ordered super-

lattices

In order to explain the relative intensity of the two states at

saturation seen in flash desorption n~~ = 21 Tamm and Schmidt7

propose a 1 2

model wherein the initial hydrogen adsorption is atomic and at B sites in a

C(2x2) pattern At full coverage the remaining B sites are filled with

hydrogen molecules B sites are chosen over C sites using molecular orbital

55arguments A sites are rejected as there are too many to satisfy this model

This model satisfies the second order kinetics observed for the ~2 state and

the first order kinetics observed for the ~l state In addition it explains

14 2 7 their saturation coverage of gt 3xl0 moleculescm One weakness of this

model pointed out by Tamm and Schmidt is that it is not obvious the model can

account for the observed LEED and work function results Also no molecular

vibrations have been observed

18

19yonehara and Schmidt propose a three bonding state model that is

more consistent with the low temperature LEED results In their model an

additional state labeled ~3 is always present at the surface in A or B

2sites with a density of 5xl014 atomscm bull This state orders on the surface at

low temperatures to give the C(2x2) pattern observed At room and higher

temperatures it goes into solution (or perhaps disorders) becoming invisible

to LEED In the type A site model adsorption from the background fills in the

remaining A sites sequentially with the ~2 and ~l states This model is similar

18to that proposed by Estrup and Anderson except that the ~3 state although

invisible to LEED at room and higher temperatures occupies 14 of the type A

sites This model eliminates the stoichiometric objection to Estrup and

Andersons model The model depends on the possibility of type A sites being

occupied by states with considerably different properties If the ~3 state

occupies type B sites ~2 state adsorption fills in the remaining type B sites

and then ~l hydrogen adsorbs into random or possibly type C sites This

second model does not explain satisfactorily work function or LEED spot pattern

experimental data

56Propst found using a point defect type calculation that only a

B-site model of adsorption would produce the observed vibrational energies of

5769 meV and 135 meV Park using LEED intensity arguments found that only

adsorption into sites with 4-fold symmetry on the (100) surface (type B or type

C sites) would be consistent with broadening and splitting of the half-order

beams while the integral beams remain sharp

8Madey and Yates offer a critique of several of the proposed models

INVESTIGATION OF HYDROGEN CHEMISORPTION ON THE TUNGSTEN (100) SURFACE

Stephen Parker Withrow PhD Department of Physics

University of Illinois at Urbana-Champaign 1975

A new apparatus for surface studies has been constructed and placed

in operation The instrument incorporates a monoenergetic beam of electrons

a high resolution high sensitivity electrostatic energy spectrometer

for electrons and ions plus other hardware and electronics necessary to

perform work function measurements Auger electron spectroscopy electron

stimulated desorption of ions low energy electron diffraction and electron

scattering energy loss measurements It has been used to investigate

geometric energetic and kinetic properties associated with the chemisorpshy

tion of hydrogen on a single crystal tungsten (100) surface over a temperashy

ture range from 600 c to -1500 C

Auger spectroscopy data indicate heating thecrystl11 to 2206degK

is sufficient to clean the surface of carbon and oxygen Howeverweakl

structure is observed between 475 eV and 520 eV Possible sources of this

structure have been considered

The work function changes observed at both 60 0 C and -1500 C as a

function of exposure of the crystal surface to hydrogen are equivalent within

experimental error The full monolayer changes in work function with hydrogen

adsorption are 098 eV and 092 eV at the two temperatures respectively

Cooling the clean surface from 2200 C to _1500 C increases the work function

~ 005 eV

Hydrogen ions are desorbed from the dosed tungsten surface following

electron impact The distribution in energy of these ions has a Gaussian

line shape with a peak intensity at an ion energy of 30 eV and a full width

at half maximum of 26 eV The ion yield is found to be coverage dependent

reaching a maximum at work function changes of 018 eV and 011 eV for desorpshy

tion at 600C and -150oC respectively The total desorption cross section

-20 2 0for H+ from the S2 hydrogen state 1S calculated to be 5xlO cm at 60 C

-19 2 0and 12xlO cm at -150 C

A new hydrogen binding state labeled the fast state has been

observed and characterized using ESD techniques This state is desorbable

if the undosed crystal is cooled below room temperature It is populated to

-4 a very low coverage ~10 monolayer by adsorption from the background of

some other t han H T e tota esorpt10n cross cm2spec1es 2 hId sect1on 10- 17 and

energy distribution of the fast state are markedly different than for the S

states

LEED pattern changes observed as a function of hydrogen coverage

are in good agreement with results published in the literature The intenshy

sity of the c(2x2) pattern associated with the S2 hydrogen state maximizes

at a work function change with adsorption of 015 eV LEED elastic intensity

profiles have been obtained for several low index beams These are compared

to theoretical microscopic model calculations and to published experimental

profiles A c(2x2) LEED pattern is obtained if the crystal is cooled below

room temperature Attempts have been made to correlate the appearance of this

pattern to changes in measurements taken in this low temperature range with

other techniques No significant bulk to surface diffusion of hydrogen was

found No evidence was found from work function and electron scattering

experiments to support the hypothesis that the low temperature c(2x2)

pattern results from hydrogen on the surface In addition the fast hydrogen

state is not believed to be responsible for the c(2x2) pattern LEED

intensity profiles from the undosed cold surface c(2x2) pattern and from the

c(2x2) pattern associated with the ~2 hydrogen state are not identical an

indication that the atomic geometry responsible for the two patterns may not

be the same

Electron scattering energy loss spectra have been obtained as a

function of hydrogen coverage for loss energies of 0 eV ~ w lt 40 eV Hydrogen

adsorption has a marked effect on the spectra Fine structure in the data

has been discussed and compared to photoemission data Cooling the clean

crystal below room temperature causes only a slight change in the energy loss

spectrum

iii

ACKNOWLEDGMENTS

I would like to sincerely thank my advisor Professor F M Propst

for guidance given throughout the course of this research and for his knowledge

of surface research which helped shape this work Dr C B Duke for many

helpful conversations and for assistance in the analysis of the data my wife

Lois for her patience and steadfast encouragement my parents and parents-inshy

law for their generous and needed support Paul Luscher and John Wendelken for

their creativity and cooperation in the design of the apparatus and for many

interesting scientific discussions Jim Burkstrand and Chris Foster for their

friendship and technical help on many occasions Bob Bales George Bouck Bob

Fults Bill Beaulin Lyle Bandy and Bill Lawrence for design assistance and

expert workmanship in the construction of the apparatus Nick Vassos and Bill

Thrasher for the imaginative and superior work in materials processing necesshy

sary in the construction of the instrument Gene Gardner and co-workers for

able assistance in electronics Bob MacFarlane and assistants for their high

quality draftmanship Jack Gladin and Jim Gray for an excellent job on the

photography needed for this research Bob Ebeling for his cooperation in the

print shop Chris Fleener for an excellent typing of the manuscript and the

rest of the staff of the Coordinated Science Laboratory whose efforts have

been gratefully appreciated during my tenure as a graduate student

bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

iv

TABLE OF CONTENTS

Chapter Page

1 MOTIVATION AND REVIEW OF LITERATURE bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 1

11 Introduction and Motivationbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 1

12 Review of Literature bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 4o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

121 Introductionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 4 122 Thermal Desorptionbull o 4 123 LEED and HEEDbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 7 124 Electron Scattering Field Emission and

Photoemissionbullbullbull 100 bullbullbullbullbull 0 bullbullbullbull

125 Electron Stimulated Desorptionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 12 126 Work Functionbullbullbullbullbullbullbullbullbull 13o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

1 27 Chemical Probes bullbullbullbullbullbullbull bullbullbullbull bullbullbullbull 14I) O

128 Chemisorption Models bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 15 1 29 Theoretical Treatment bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 19 1 210 Literature Surnmaryo 20

13 Review of Experimental Techniques bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 21

2 EXPERIMENTAL APPARATUS bullbullbullbullbullbullbullbullbullbullbullbull ~ bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 22

21 CGeneral Description bull 22

22 Vacuum System and Magnetic Shieldingbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 22

23 Electron Gun and Gun Electronics bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 25

24 Energy Spectrometer 29

25 Spectrometer Electronics bullbull 45o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

26 LEED Display System 490 bullbullbull 0 bullbullbullbullbullbullbull 0 bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull I)

27 Crystal Manipulator Assembly bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 51

28 Crystal Preparation and Cleaning Procedures bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 57

29 Gas Handling System o bullbullbullbullbullbullbullbullbullbullbullbullbull o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 62

210 Mass Spectrometer bullbullbullbull o 64

3 EXPERIMENTAL PROCEDURE AND RESULTS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 65

31 Introductionbullbullbullbullbullbullbullbullbullbullbullbull 65

32 Experimental Procedure o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull o bullbullbullbullbullbullbull 65 321 Work Function and Coverage Measurements bullbullbullbullbullbullbullbullbullbullbullbullbull 65 322 Primary Electron Beam Energy and Current

Measurements 6811 bullbullbullbullbullbullbullbullbullbullbullbullbullbullbull bullbullbullbull

323 Target Temperature Measurement bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 69 324 Electron Scattering and Auger Measurements bullbullbullbullbullbullbullbullbullbull 69 325 Ion Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 72o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

v

33 Ion Species Identificationbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 74

34 Work Function Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 77

35 u+ Ion Desorption Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 83 351 ut Ion Energy Distributionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 83 352 Desorption Curve Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 93 353 Temperature Dependence of the Fast State a+ Ion

Signal 4) 96

354 Additional Measurements on the Fast State bullbullbullbullbullbullbullbullbullbull 98

36 Low Energy Electron Diffraction Data bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull lll

37 Electron Scattering Data bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 119

4 DISCUSSION OF RESULTS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 126

41 Auger Electron Spectroscopy Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 126

42 Work Function Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 127

43 Electron Stimulated Desorption Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 129 431 Room Temperature Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 130 432 Low Temperature Fast State Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 135 433 Low Temperature Dosed Surface Results bullbullbullbullbullbullbullbullbullbullbullbullbull142

44 Low Energy Electron Diffraction Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 145

45 Electron Scattering Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 161

5 SUMMARY AND CONCLUSIONS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull165

REFERENCES 171

APPENDIX A 0 176 poundI

APPENDIX B 9 c 179

APPENDIX C 0 182 0 bull

VITA 190lilt bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull bullbull

1

CHAPTER 1

MOTIVATION AND REVIEW OF LITERATURE

11 Introduction and Motivation

The experimental investigation of chemisorption of gases onto metallic

surfaces has uncovered in the past decade a baffling complexity of adsorption

geometries and binding state kinetics and energetics A growing interest in

surface phenomena generated by the questions raised in these researches and

by the long anticipated use of the answers has led to the rapid development

of new surface probing techniques and to an increase in the availability of new

technologies compatible with surface research Aided by the new technologies

and techniques chemisorption research has been undertaken on many different

systems (system =solid substrate + gas adsorbate) under a wide range of

environmental conditions Surface theoreticians have proposed microscopic and

phenomenological models with which data have been analyzed and compared A

large experimental and theoretical literature in the field has resulted

The study of adsorption on metal surfaces has among its primary

objectives a detailed description of the physical geometry of the atoms in the

surface layers and a quantitative understanding of electronic and chemical

surface phenomena While considerable progress has been made in documenting

gas-surface phenomena it has been only in the past three years that any

credible claims to understanding specific surface overlayer geometries have

1been made At the same time many basic questions involving system kinetics

and energetics have not been adequately resolved 2

It has become apparent from the failure of the research effort to

satisfy the above objectives and from seemingly conflicting data that adequate

2

characterization is basic to all surface experimentation The need for such

characterization has been strongly emphasized by disagreement between quasishy

similar research performed at different locations Small impurity concentrashy

tions small changes in crystal orientation minor temperature differences etc

often have considerable effects on experimental measurements

A corollary to the necessity of being able to characterize the

surface is the importance of being able to control the surface characterization

In order to compare data it is necessary to reproduce all experimental paramshy

eters accurately The theory to which experimental data are compared is

constrained by complexity to be based on ideal clean substrates with a

definable gas-surface interaction It is desirable to be able to choose the

adsorbate-substrate system and experimental parameters that can also be used

with theoretical models Of course compromises are inevitable

One approach in surface research has been to incorporate several

experimental techniques into one apparatus At the present time this approach

is increasingly more common A combination of surface sensitive probes offers

many advantages It provides for more checks of the data against results

published in the literature It allows for better characterization of the

experimental parameters It allows surface phenomena to be probed in more than

one way in order to eliminate ambiguities andor to reinforce conclusions It

eliminates the uncertainties that arise from data taken at different locations

using the techniques separately hence it allows better correlations between

data taken using the different techniques

In this study several aspects of the chemisorption interaction of

hydrogen with the low index (100) plane of single crystal tungsten denoted

3

W(100) have been investigated The choice of the tungsten-hydrogen system is

motivated by several factors The literature raises questions on the unusual

state behavior observed when hydrogen adsorbs on W(100) questions which our

apparatus is especially well-suited to study In addition the interaction of

hydrogen with close packed planes of single crystal metals including W(100)

is of particular interest to theoreticians since hydrogen is the simplest

molecular gas as detailed in Sections 128 and 129 below recent theories

of chemisorption have been applied to the hydrogen-tungsten system Tungsten

was chosen as the substrate metal because of its relative ease of preparation

and cleaning and because the large body of literature on the hydrogen-tungsten

system indicates wide interest in this interaction

In particular this research is concentrated in three areas Tung~

sten when cleaned and subsequently cooled to below room temperature exhibits

an anomalous centered (2x2) denoted C(2x2) low energy electron diffraction

pattern The possibility that this is caused by hydrogen diffusing from the

bulk or adsorbing from the background has been investigated Secondly the

room temperature and low temperature hydrogen adsorption binding states have

been studied using electron stimulated desorption and electron spectroscopy

Finally low energy electron diffraction has been used in an investigation of

the atomic geometry of the clean tungsten surface and the geometry of one of

the hydrogen binding states

The remaining sections of this chapter summarize literature relating

to this research Chapter 2 details the construction and performance of the

experimental apparatus A presentation of the experimental data is made in

Chapter 3 A discussion of the results is made in Chapter 4 A summaryof

4

the research and conclusions from it are given in Chapter 5

12 Review of Literature

121 Introduction

Experimental investigation of the adsorption of hydrogen on tungsten

has been carried out using field emitter tips macroscopic polycrystalline

surfaces and all the low index orientation single crystal faces including

W(lOO) From the wide range of results it is clear that the substrate exerts

a strong influence on the adsorption properties The question has been raised

whether adsorption models postulated from polycrystalline substrates can be

3extended even qualitatively to single crystal systems Likewise there are

such sharp differences in the chemisorption behavior of hydrogen onto the

different single crystal tungsten surfaces that the specific physical process

governing adsorption on each face appears to be unique to that face This

review will therefore be concerned for the most part with literature dealing

with hydrogen adsorption on W(lOO) the gas-substrate system used in this

research

The following data review is presented in subsections each concenshy

trating on a specific experimental technique or class of experiments Subsecshy

tion 127 includes some gas mixing literature which was considered relevant

and Subsection 129 gives a bibliography of some related chemisorption

theoretical papers

122 Thermal Desorption

One technique widely employed in investigation of the kinetics of gas

456adsorption has been thermal or flash desorption By monitoring the system

5

pressure changes of a species while rapidly heating the crystal inferences

can be made on the existence of binding states adsorption energies and reaction

7 8 order Heating is accomplished by a focussed light beam electron bombardshy

7 9ment or ohmic losses bull

The strongest indication that hydrogen exhibits multiple state

adsorption behavior on W(lOO) comes from flash desorption data Tamm and

8Schmidt 7 and Madey and Yates found for room temperature adsorption that two

large peaks are observed in the plot of pressure of H2 versus temperature

These peaks are centered at 4500 K and approximately 550oK Only hydrogen

molecules are observed to be desorbed from the surface Using desorption rate

456theory the lower temperature (~l) state is found to obey first order

kinetics the peak temperature being independent of initial coverage The

higher temperature (~2) state obeys second order kinetics the peak temperature

moving down from approximately 5800 K to 5500 K as the initial coverage is

increased

In addition to the two main peaks several small peaks comprising a

total of less than 10 percent of the desorbed hydrogen are observed during a

flash when the crystal is cooled to nitrogen temperature before hydrogen

exposure These states are believed to result from adsorption onto edges and

other non-(lOO) crystal orientations 8

The dependence of the desorption energy Ed on coverage is weak

~l 8 ~2 8with Ed = 252 kcallmole and Ed = 323 kcalmole As would be predicted

from this result the ~2 state is largely populated before the ~l state begins

to fill The ratio of the populations in the states at full coverage n~1n~2

8 lOis 2111 Jelend and Menzel point out that the cross section for electron

6

desorption (see Subsection 125) from the ~1 state is considerably smaller

than that of the ~2 state a result not predicted from the experimental flash

desorption binding energies

11Tibbetts using a stepped temperature flash desorption technique

has shown that hydrogen is bound to the surface of po1ycrysta11ine tungsten

W(110) and W(lll) with a distribution of binding energies ranging from

162 kca1mo1e (07 eV) to 415 kca1mo1e (lS eV) Using this technique

desorption at a constant temperature from a distribution of states can be easily

distinguished from desorption from a state with a unique binding energy While

he did not extend his work to W(100) these results suggest that hydrogen does

not bind to the surface at a few energies but rather with a range of binding

energies Tibbetts also shows that the determination of the order of the

desorption process is difficult for adsorption into a range of binding energies

In experiments using a combination of hydrogen and deuterium as room

temperature adsorbates the elemental composition of the desorbed ~1 and ~2

states shows complete isotopic mixing between the states7S independent of the

adsorption sequence or ratios However for low temperature adsorption

complete state conversion is not observed for all the additional small peaks

Where mixing is observed atomic adsorption is suggested Where mixing is not

observed molecular adsorption is inferred

Some theoretical models indicate that multiple peak flash desorption

spectra may be due to binding energy changes caused by adsorbate-adsorbate

interactions12 or order-disorder transactions during f1ash3 13 rather than

actual distinct multiple binding states under the adsorption conditions

7

123 LEED and HEED

Low energy electron diffraction or LEED refers to the coherent

scattering of a monoenergetic focused electron beam from the surface of a

crystalline solid Electrons in the energy range 0 lt E lt 300 eV interact

strongly with the ion cores and valence electrons of the solid The inelastic

mean free path is on the order of 10 X As a consequence electrons which are

observed elastically have been backscattered within the first several surface

layers in the solid making LEED a surface sensitive probe 14 15 The periodic

translational symmetry of the surface leads to the momentum conservation laws

I

= (11)~I ~II + g

= (2mEh2)t sin reg (12)~II

where ~II is the momentum parallel to the surface of the emerging electron and

~ is the parallel momentum of the incident electron The vector g is a

reciprocal lattice vector of the solid surface and reg is the polar angle of the

incident beam As a result of the conservation laws the scattered electrons

emerge in a series of beams called a spot pattern which is a manifestation

of the surface symmetry The beams are labelled by their corresponding g

vectors Information on the actual configuration of the surface ion cores is

extracted from the plot of elastically scattered electron intensity versus

16 17incident energy called elastic intensity profiles or I-V curves

Changes in the LEED spot pattern resulting from hydrogen adsorption

18onto W(lOO) at room temperature have been reported by Estrup and Anderson and

19by Yonehara and Schmidt As hydrogen is adsorbed at room temperature the

8

(lxl) pattern characteristic of clean tungsten changes into a weak C(2x2)

pattern at a hydrogen coverage e of less than 005 monolayer (where e = 1

15 2monolayer at a coverage of 2xlO hydrogen atomsem which is 2 hydrogen atoms

for each surface tungsten atom) The (1212) spots become sharp and intense

with increased hydrogen adsorption reaching a maxima in intensity at approxishy

19 18mately 015 monolayer or e = 025 monolayer (Coverage in both cases was

determined by flash desorption) The (1212) spots split into four new spots

which move apart continuously and upon further hydrogen adsorption eventually

19form 13-order streaks at a coverage of ~ 035 monolayer These in turn

change at higher coverage into poorly developed 14-order features with the

(012) spots missing Finally near e = 10 all extra diffraction features

are gone and the LEED pattern returns to a (lxl) symmetry This series of LEED

pattern changes is reversible with coverage

For hydrogen adsorption at 1950 K the C(2x2) pattern develops

maximizes in intensity at e ~ 18 and splits into four beams similar to room

temperature adsorption However the higher coverage LEED structures third

19 0order streaks and fourth order spots are not observed At 78 K adsorption

of hydrogen produces half-order C(2x2) spots that maximize in intensity at

e ~ 027 Additional hydrogen adsorption beyond this causes the half-order

19spot intensity to decrease but with no splitting of the spots

If clean W(lOO) is allowed to cool below room temperature a sharp

19and intense C(2x2) pattern is observed even though total coverage due to the

background of H2 and CO is less than e = 001 as determined by flash desorption

measurements The intensity of these half-order beams can be altered by

additional hydrogen adsorption or by various hydrogen and oxygen heat treatmenm

9

l9Yonehara and Schmidt have postulated that the extra spots may result from

hydrogen endothermally soluble in tungsten diffusing to the surface Possible

alternative explanations include the reconstruction of the clean tungsten

surface or impurity over layer adsorption from the background It should be

20noted that Plummer and Bell saw no effects upon cooling their field emitter

tip to 78 0 K that they could attribute to hydrogen diffusion to the surface

Elastic intensity profiles for several beams have been measured by

18 21Estrup and Anderson and by Papageorgopoulos and Chen at room temperature

Estrup and Anderson reported profiles for (00) and (1212) tEED lJeams out

to 300 eV The intensity variation for the (OO) beam was found to be indepenshy

dent of hydrogen coverage a manifestation of the relatively weak scattering

cross section of a hydrogen ion core relative to that of tungsten The

additional beams observed on adsorption would be expected to be more sensitive

13to the adsorbed overlayer Estrup found the intensity of the (1212) beams

for H W(100) and D W(100) to decrease as a function of temperature between2 2

300o

K and 6000 K at a constant incident energy but neither obeyed the Debyeshy

22Waller relation Estrup interpreted this result to imply the adsorbed

species are directly involved in the diffraction process

19Yonehara examined the tEED spot intensities as a function of

temperature down to liquid nitrogen temperatures His results are examined

in the data discussion in Chapter 4

The hydrogen-WlOO) system has also been studied using high energy

electron diffraction or HEED23 This technique utilizes the strong forward

scattering of high energy electrons in probing the surface by having the

incident beam strike the solid at grazing incidence The resulting diffraction

10

patterns are sensitive to the translational symmetry of the surface As in

LEED a diffraction pattern with C(2x2) symmetry is observed during initial

hydrogen adsorption With further hydrogen exposure the beams broaden some

splitting is observed and eventually the half-order intensity decreases

completely The width of the observed half-order beams has been interpreted to

indicate hydrogen initially adsorbs into a large number of small square or

rectangular domains each consisting of 20 to 25 atoms with sides parallel to

the [100J directions

124 Electron Scattering Field Emission and Photoemission

The electronic structure of the surface layers of solids can be

characterized by electron emission spectroscopies Such techniques require a

surface sensitive perturbation on the gas-solid system leading to an electron

yield A spectrometer is used to energy analyze the emitted or scattered

24 25electrons In electron scatter~ng a monoenergetic beam of electrons is

directed onto a crystal The inelastically scattered signal containing

quantitative information about electronic excitations at the surface is

measured as a function of energy loss Field emission is a technique in which

application of a high positive electric field to the surface lowers the potenshy

26tia1 barrier for electrons to tunnel out of the system The total energy

distribution of electrons emitted as a result of the applied field contains

bull 27 28spectroscopic data on the adsorbed spec~es Structure induced by the

adsorbate in the elastic tunneling spectrum is thought to be a measure of the

d f i h f h d b 2930e1ectron~c ens~ty 0 states n t e v~c~n~ty 0 tea sor ate The

inelastic tunneling spectrum reveals discrete excitation energies in the

11

gas-surface system In photoemission or ftotoelectron spectroscopy an incident

monochromatic light beam incident on the solid induces emission of electrons by

interband transitions with or without an inelastic scattering after the initial

31 exc i tat~on

The vibrational spectrum of hydrogen adsorbed on the tungsten (100)

32surface has been observed by Propst and Piper using inelastic low energy

20electron scattering and by Plummer and Bell using field emission Propst

and Piper found two peaks in the energy loss distribution E - E pr~mary scattered

at 135 meV and 69 meV for all coverages used up to saturation These results

are interpreted as indicating hydrogen atoms are multiply bonded to tungsten

33atoms No losses near 550 meV the vibrational energy of molecular hydrogen

were observed Plummer and Bell measured the total energy distribution of

field emitted electrons from tungsten (100) as a function of coverage of

hydrogen and deuterium at several temperatures Discrete losses at 100 meV

and 50 meV for deuterium which correspond to the 135 meV and 69 meV vibrational

losses observed by Propst and Piper were evident for deuterium in the a state2

No discrete energy losses at 550 meV or 390 meV the molecular vibrational

levels of H2 and D2 respectively were measured Hence there was no observable

molecular adsorption at room temperature

Additional structure in the field emission total energy distribution

is observed as a function of hydrogen adsorption The clean surface spectrum

has two peaks 04 eV and 15 eV down from the Fermi level which have been

d ifi d h h d f d h h f 34 A h d ~ ent e as ~g ens~ty 0 states assoc~ate w~t t e sur ace s y rogen

(or deuterium) coverage increases between 0 and 5xl014 atomscm2

ie while

the ~2 state is filling the surface states disappear and the field emitted

12

energy distribution changes from that of clean tungsten to a spectrum indicative

of resonance tunneling through a level positioned 09 eV below the Fermi energy

There is a shoulder at -11 eV and an energy dependent background With

further adsorption up to saturation this energy distribution characteristic

of the ~2 state shifts to a broad level centered at 26 eV below the Fermi

level This is interpreted to indicate that the ~1 state does not adsorb on

top of the ~2 state

35The photoemission results of Wac1awski and P1ummer and Feuerbacher

36and Fitton are in agreement with those obtained using the field emission

technique As hydrogen adsorbs on the surface emission from the 04 eV

surface state decreases The level 09 eV down from the Fermi energy as well

as a level 5 eV down remain unchanged by adsorption At full coverage a strong

peak at -25 eV is observed

125 Electron Stimulated Desorption

Desorption of energetic ions and neutrals from a gas over layer can

be induced by electron bombardment This process termed electron stimulated

desorption or ESD37383940 is described below in the discussion of the data

in Chapter 4 In brief intensity and energy measurements of the ion yield can

be interpreted in terms of the parameters characterizing the binding energy

curve of the adsorbed state and adsorption states can be classified by cross

section for desorption_ ion energy distributions primary energy threshold for

desorption etc

ESD has been employed to study adsorption kinetics for the hydrogenshy

41 ~ tungsten (100) system Madey obtained a coverage dependence yield of H ions

13

observing an ion signal during the initial stages of hydrogen adsorption that

reached a maximum at a coverage of e = 017 and then decayed to zero yield

23 2after e = 075 the maximum cross section obtained was 18x10- cm at a

primary electron energy of 100 eV Tamm and Schmidt7 reported a maximum cross

section for H+ desorption of 10- 21 cm 2bull Benninghoven Loebach and Treitz42

10and Je1end and Menze1 obtained qualitatively similar (to Madey) coverage

dependencies on tungsten ribbons which were believed to be predominantly of

(100) crystallites the main difference between these results and Madeys is

the observation of a residual signal at high hydrogen coverage

The cross section results indicate that the a state even though it2

is more tightly bound than the a state provides more hydrogen ions upon1

electron impact The cross section for ion desorption from the a1 state is

d41 10-25est1mate to be 1ess t han cm2 bull

+No energy distributions of the desorbed H ions have been reported

The dependence of the desorption cross section on the primary electron energy

10has not been investigated Je1end and Menzel found their results independent

of temperature for temperatures above room temperature No ESD results for

otemperatures less than 300 K have been reported

126 Work Function

The work function is the minimum amount of energy required to remove

43 an electron from a meta1 An adsorbate may alter the potential barrier for

electrons at the surface and hence change the work function Such is the case

for hydrogen on W(100) Hydrogen adsorbs as an electronegative ion on W(100)

thereby increasing the work function

14

44Using a retarding potential current-voltage techn~que Madey and

8Yates reported a linear relation between work function change and hydrogen

coverage over the entire coverage range for adsorption occurring at a

temperature 0 f 2= 300 0 K Coverages were determined by flash desorption

Total work function change at one monolayer was found to be 090 eV which

corresponds to a constant dipole moment per adsorbed atom of 015 Debye

15 2(assuming a total coverage of 15xlO atomsem)

Estrup and Anderson18 using the retarding potential technique and

Plummer and Bel120 using a field emission Fowler-Nordheim plot45 found a

work function change versus coverage relationship which deviated only slightly

8from being linear in near agreement with the results of Madey and Yates

46 47Becker using field emission and Armstrong using a retarding potential

technique found a work function change for the H W(100) system of approximately2

48090 eV when e = 10 Hopkins and Pender obtained an anomalously low value

49of 054 eV while Collins and Blott obtained a high value of 109 eV for the

same coverage

127 Chemical Probes

Hydrogen adsorption has also been studied using a chemical probe

ie by coadsorbing some additional different gas with the hydrogen and

observing the effect this additional species had on results obtained using

techniques discussed in previous sections

50Yates and Madey found using flash desorption that CO coadsorption

onto a hydrogen monolayer displaces both the Sl and S2 binding states with

near unity probability in doing so initially replacing these states with a

15

less tightly bound hydrogen state After high CO exposure no hydrogen remains

on the surface if the coadsorption is done at room temperature At lower

temperatures only about 10 percent of the hydrogen is desorbed Jelend and

Menzel lO 5l found that the ionic and total desorption cross sections for

hydrogen the sticking probability and bond strength of hydrogen are all

affected by preadsorption of CO or oxygen Yates and Madey3 found that N2

slowly replaces preadsorbed hydrogen at temperatures greater than 300oC Below

2730

K the displacement of hydrogen by N2 does not occur

These results all show that small impurity concentrations of CO N

and 0 if intentionally introduced do affect the resulting hydrogen chemishy

sorption properties These experiments underscore the necessity of surface

chemical characterization

128 Chemisorption Models

7 18 19 52 Several models have been proposed to account for the

experimental results from hydrogen chemisorption on W(lOO) Each of these

models embodies some combination of the following three phenomenological

assumptions 1) The tungsten surface is ideal planar and free of contamishy

nants 2) There are available two types of adsorbates namely hydrogen atoms

53and molecules 3) There are three high symmetry logical bonding sites on

a BCC (100) face usually designated by the letters A B and C These

positions are A a bridge site with 2 nearest neighbors and the hydrogen

equidistant from two adjacent surface tungsten atoms B a four fold symmetric

site with 5 nearest neighbors and the hydrogen located on a surface normal that

passes through the center of a surface primitive unit cell and C the hydrogen

16

located directly above a surfac~ atom with one nearest neighbor There are

15 152x10 A sitescm2 and 1x10 B ore sites~m2 on middotW(100)

The proposed models attempt to explain or at least be consistent

with empirical results presented in Sections 121 through 127 A summary

of salient features of the reported research includes a) a linear work function

15change with coverage b) a saturation coverage of between 15 and 20x10 H

2atomscm c) a coverage dependent adsorption layer symmetry including a we11shy

ordered surface at low coverages d) the appearance at low temperatures of a

surface translational symmetry not equivalent to room temperature clean

tungsten e) adsorption into states which produces a two state thermal desorpshy

tion spectrum with relative coverages in the two states of nQ n = 21 f) a ~1 ~2

coverage dependence for the observed electronic structure including the

absence at high coverage of structure observed at low coverages and g) no

observed vibrational levels which correspond to free H2 molecule

Following is a short review of several proposed surface models

18Estrup and Anderson account for their LEED spot pattern results

using a model wherein the hydrogen molecules dissociate upon adsorption and

migrate initially to form islands of hydrogen atoms located at type A sites in

a e(2x2) structure Dissociation is assumed since the e(2x2) pattern maximizes

15 2at a coverage of 25x10 mo1ecu1escm whereas without dissociation it would

maX1m1ze at 5 1015 mo ecu1es cm2bull e site adsorptionx 1 Although type B or type

could also produce the initial e(2x2) LEED pattern these sites are rejected

15 2because of the following 1) There are only 1x10 B or e sitescm This is

not enough sites to accommodate a full monolayer coverage if all binding is

atomic and if it is into equivalent sites as suggested by the linear work

17

function change with coverage 2) While a complete 14-order LEED pattern is

possible with type B or type C sites it is not possible to produce a 14-order

54pattern with (012) spots missing using B or C sites As the adsorption

14 2proceeds past a surface density of 5xlO hydrogen atomscm the adatoms conshy

tinuously rearrange into long period superlattices with antiphase domains

that produce successively the 12-order spot splitting the 13-order streak

(only partly ordered domains) the 14-order pattern and finally the (lxl)

full coverage structure

This model has been criticized for not accounting for the two

observed binding states and for predicting a maximum in intensity of the half

order spots at e = 025 these beams maximize at less than that coverage In

addition the authors themselves questioned the idea of well-ordered super-

lattices

In order to explain the relative intensity of the two states at

saturation seen in flash desorption n~~ = 21 Tamm and Schmidt7

propose a 1 2

model wherein the initial hydrogen adsorption is atomic and at B sites in a

C(2x2) pattern At full coverage the remaining B sites are filled with

hydrogen molecules B sites are chosen over C sites using molecular orbital

55arguments A sites are rejected as there are too many to satisfy this model

This model satisfies the second order kinetics observed for the ~2 state and

the first order kinetics observed for the ~l state In addition it explains

14 2 7 their saturation coverage of gt 3xl0 moleculescm One weakness of this

model pointed out by Tamm and Schmidt is that it is not obvious the model can

account for the observed LEED and work function results Also no molecular

vibrations have been observed

18

19yonehara and Schmidt propose a three bonding state model that is

more consistent with the low temperature LEED results In their model an

additional state labeled ~3 is always present at the surface in A or B

2sites with a density of 5xl014 atomscm bull This state orders on the surface at

low temperatures to give the C(2x2) pattern observed At room and higher

temperatures it goes into solution (or perhaps disorders) becoming invisible

to LEED In the type A site model adsorption from the background fills in the

remaining A sites sequentially with the ~2 and ~l states This model is similar

18to that proposed by Estrup and Anderson except that the ~3 state although

invisible to LEED at room and higher temperatures occupies 14 of the type A

sites This model eliminates the stoichiometric objection to Estrup and

Andersons model The model depends on the possibility of type A sites being

occupied by states with considerably different properties If the ~3 state

occupies type B sites ~2 state adsorption fills in the remaining type B sites

and then ~l hydrogen adsorbs into random or possibly type C sites This

second model does not explain satisfactorily work function or LEED spot pattern

experimental data

56Propst found using a point defect type calculation that only a

B-site model of adsorption would produce the observed vibrational energies of

5769 meV and 135 meV Park using LEED intensity arguments found that only

adsorption into sites with 4-fold symmetry on the (100) surface (type B or type

C sites) would be consistent with broadening and splitting of the half-order

beams while the integral beams remain sharp

8Madey and Yates offer a critique of several of the proposed models

Hydrogen ions are desorbed from the dosed tungsten surface following

electron impact The distribution in energy of these ions has a Gaussian

line shape with a peak intensity at an ion energy of 30 eV and a full width

at half maximum of 26 eV The ion yield is found to be coverage dependent

reaching a maximum at work function changes of 018 eV and 011 eV for desorpshy

tion at 600C and -150oC respectively The total desorption cross section

-20 2 0for H+ from the S2 hydrogen state 1S calculated to be 5xlO cm at 60 C

-19 2 0and 12xlO cm at -150 C

A new hydrogen binding state labeled the fast state has been

observed and characterized using ESD techniques This state is desorbable

if the undosed crystal is cooled below room temperature It is populated to

-4 a very low coverage ~10 monolayer by adsorption from the background of

some other t han H T e tota esorpt10n cross cm2spec1es 2 hId sect1on 10- 17 and

energy distribution of the fast state are markedly different than for the S

states

LEED pattern changes observed as a function of hydrogen coverage

are in good agreement with results published in the literature The intenshy

sity of the c(2x2) pattern associated with the S2 hydrogen state maximizes

at a work function change with adsorption of 015 eV LEED elastic intensity

profiles have been obtained for several low index beams These are compared

to theoretical microscopic model calculations and to published experimental

profiles A c(2x2) LEED pattern is obtained if the crystal is cooled below

room temperature Attempts have been made to correlate the appearance of this

pattern to changes in measurements taken in this low temperature range with

other techniques No significant bulk to surface diffusion of hydrogen was

found No evidence was found from work function and electron scattering

experiments to support the hypothesis that the low temperature c(2x2)

pattern results from hydrogen on the surface In addition the fast hydrogen

state is not believed to be responsible for the c(2x2) pattern LEED

intensity profiles from the undosed cold surface c(2x2) pattern and from the

c(2x2) pattern associated with the ~2 hydrogen state are not identical an

indication that the atomic geometry responsible for the two patterns may not

be the same

Electron scattering energy loss spectra have been obtained as a

function of hydrogen coverage for loss energies of 0 eV ~ w lt 40 eV Hydrogen

adsorption has a marked effect on the spectra Fine structure in the data

has been discussed and compared to photoemission data Cooling the clean

crystal below room temperature causes only a slight change in the energy loss

spectrum

iii

ACKNOWLEDGMENTS

I would like to sincerely thank my advisor Professor F M Propst

for guidance given throughout the course of this research and for his knowledge

of surface research which helped shape this work Dr C B Duke for many

helpful conversations and for assistance in the analysis of the data my wife

Lois for her patience and steadfast encouragement my parents and parents-inshy

law for their generous and needed support Paul Luscher and John Wendelken for

their creativity and cooperation in the design of the apparatus and for many

interesting scientific discussions Jim Burkstrand and Chris Foster for their

friendship and technical help on many occasions Bob Bales George Bouck Bob

Fults Bill Beaulin Lyle Bandy and Bill Lawrence for design assistance and

expert workmanship in the construction of the apparatus Nick Vassos and Bill

Thrasher for the imaginative and superior work in materials processing necesshy

sary in the construction of the instrument Gene Gardner and co-workers for

able assistance in electronics Bob MacFarlane and assistants for their high

quality draftmanship Jack Gladin and Jim Gray for an excellent job on the

photography needed for this research Bob Ebeling for his cooperation in the

print shop Chris Fleener for an excellent typing of the manuscript and the

rest of the staff of the Coordinated Science Laboratory whose efforts have

been gratefully appreciated during my tenure as a graduate student

bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

iv

TABLE OF CONTENTS

Chapter Page

1 MOTIVATION AND REVIEW OF LITERATURE bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 1

11 Introduction and Motivationbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 1

12 Review of Literature bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 4o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

121 Introductionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 4 122 Thermal Desorptionbull o 4 123 LEED and HEEDbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 7 124 Electron Scattering Field Emission and

Photoemissionbullbullbull 100 bullbullbullbullbull 0 bullbullbullbull

125 Electron Stimulated Desorptionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 12 126 Work Functionbullbullbullbullbullbullbullbullbull 13o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

1 27 Chemical Probes bullbullbullbullbullbullbull bullbullbullbull bullbullbullbull 14I) O

128 Chemisorption Models bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 15 1 29 Theoretical Treatment bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 19 1 210 Literature Surnmaryo 20

13 Review of Experimental Techniques bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 21

2 EXPERIMENTAL APPARATUS bullbullbullbullbullbullbullbullbullbullbullbull ~ bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 22

21 CGeneral Description bull 22

22 Vacuum System and Magnetic Shieldingbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 22

23 Electron Gun and Gun Electronics bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 25

24 Energy Spectrometer 29

25 Spectrometer Electronics bullbull 45o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

26 LEED Display System 490 bullbullbull 0 bullbullbullbullbullbullbull 0 bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull I)

27 Crystal Manipulator Assembly bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 51

28 Crystal Preparation and Cleaning Procedures bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 57

29 Gas Handling System o bullbullbullbullbullbullbullbullbullbullbullbullbull o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 62

210 Mass Spectrometer bullbullbullbull o 64

3 EXPERIMENTAL PROCEDURE AND RESULTS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 65

31 Introductionbullbullbullbullbullbullbullbullbullbullbullbull 65

32 Experimental Procedure o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull o bullbullbullbullbullbullbull 65 321 Work Function and Coverage Measurements bullbullbullbullbullbullbullbullbullbullbullbullbull 65 322 Primary Electron Beam Energy and Current

Measurements 6811 bullbullbullbullbullbullbullbullbullbullbullbullbullbullbull bullbullbullbull

323 Target Temperature Measurement bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 69 324 Electron Scattering and Auger Measurements bullbullbullbullbullbullbullbullbullbull 69 325 Ion Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 72o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

v

33 Ion Species Identificationbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 74

34 Work Function Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 77

35 u+ Ion Desorption Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 83 351 ut Ion Energy Distributionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 83 352 Desorption Curve Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 93 353 Temperature Dependence of the Fast State a+ Ion

Signal 4) 96

354 Additional Measurements on the Fast State bullbullbullbullbullbullbullbullbullbull 98

36 Low Energy Electron Diffraction Data bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull lll

37 Electron Scattering Data bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 119

4 DISCUSSION OF RESULTS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 126

41 Auger Electron Spectroscopy Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 126

42 Work Function Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 127

43 Electron Stimulated Desorption Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 129 431 Room Temperature Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 130 432 Low Temperature Fast State Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 135 433 Low Temperature Dosed Surface Results bullbullbullbullbullbullbullbullbullbullbullbullbull142

44 Low Energy Electron Diffraction Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 145

45 Electron Scattering Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 161

5 SUMMARY AND CONCLUSIONS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull165

REFERENCES 171

APPENDIX A 0 176 poundI

APPENDIX B 9 c 179

APPENDIX C 0 182 0 bull

VITA 190lilt bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull bullbull

1

CHAPTER 1

MOTIVATION AND REVIEW OF LITERATURE

11 Introduction and Motivation

The experimental investigation of chemisorption of gases onto metallic

surfaces has uncovered in the past decade a baffling complexity of adsorption

geometries and binding state kinetics and energetics A growing interest in

surface phenomena generated by the questions raised in these researches and

by the long anticipated use of the answers has led to the rapid development

of new surface probing techniques and to an increase in the availability of new

technologies compatible with surface research Aided by the new technologies

and techniques chemisorption research has been undertaken on many different

systems (system =solid substrate + gas adsorbate) under a wide range of

environmental conditions Surface theoreticians have proposed microscopic and

phenomenological models with which data have been analyzed and compared A

large experimental and theoretical literature in the field has resulted

The study of adsorption on metal surfaces has among its primary

objectives a detailed description of the physical geometry of the atoms in the

surface layers and a quantitative understanding of electronic and chemical

surface phenomena While considerable progress has been made in documenting

gas-surface phenomena it has been only in the past three years that any

credible claims to understanding specific surface overlayer geometries have

1been made At the same time many basic questions involving system kinetics

and energetics have not been adequately resolved 2

It has become apparent from the failure of the research effort to

satisfy the above objectives and from seemingly conflicting data that adequate

2

characterization is basic to all surface experimentation The need for such

characterization has been strongly emphasized by disagreement between quasishy

similar research performed at different locations Small impurity concentrashy

tions small changes in crystal orientation minor temperature differences etc

often have considerable effects on experimental measurements

A corollary to the necessity of being able to characterize the

surface is the importance of being able to control the surface characterization

In order to compare data it is necessary to reproduce all experimental paramshy

eters accurately The theory to which experimental data are compared is

constrained by complexity to be based on ideal clean substrates with a

definable gas-surface interaction It is desirable to be able to choose the

adsorbate-substrate system and experimental parameters that can also be used

with theoretical models Of course compromises are inevitable

One approach in surface research has been to incorporate several

experimental techniques into one apparatus At the present time this approach

is increasingly more common A combination of surface sensitive probes offers

many advantages It provides for more checks of the data against results

published in the literature It allows for better characterization of the

experimental parameters It allows surface phenomena to be probed in more than

one way in order to eliminate ambiguities andor to reinforce conclusions It

eliminates the uncertainties that arise from data taken at different locations

using the techniques separately hence it allows better correlations between

data taken using the different techniques

In this study several aspects of the chemisorption interaction of

hydrogen with the low index (100) plane of single crystal tungsten denoted

3

W(100) have been investigated The choice of the tungsten-hydrogen system is

motivated by several factors The literature raises questions on the unusual

state behavior observed when hydrogen adsorbs on W(100) questions which our

apparatus is especially well-suited to study In addition the interaction of

hydrogen with close packed planes of single crystal metals including W(100)

is of particular interest to theoreticians since hydrogen is the simplest

molecular gas as detailed in Sections 128 and 129 below recent theories

of chemisorption have been applied to the hydrogen-tungsten system Tungsten

was chosen as the substrate metal because of its relative ease of preparation

and cleaning and because the large body of literature on the hydrogen-tungsten

system indicates wide interest in this interaction

In particular this research is concentrated in three areas Tung~

sten when cleaned and subsequently cooled to below room temperature exhibits

an anomalous centered (2x2) denoted C(2x2) low energy electron diffraction

pattern The possibility that this is caused by hydrogen diffusing from the

bulk or adsorbing from the background has been investigated Secondly the

room temperature and low temperature hydrogen adsorption binding states have

been studied using electron stimulated desorption and electron spectroscopy

Finally low energy electron diffraction has been used in an investigation of

the atomic geometry of the clean tungsten surface and the geometry of one of

the hydrogen binding states

The remaining sections of this chapter summarize literature relating

to this research Chapter 2 details the construction and performance of the

experimental apparatus A presentation of the experimental data is made in

Chapter 3 A discussion of the results is made in Chapter 4 A summaryof

4

the research and conclusions from it are given in Chapter 5

12 Review of Literature

121 Introduction

Experimental investigation of the adsorption of hydrogen on tungsten

has been carried out using field emitter tips macroscopic polycrystalline

surfaces and all the low index orientation single crystal faces including

W(lOO) From the wide range of results it is clear that the substrate exerts

a strong influence on the adsorption properties The question has been raised

whether adsorption models postulated from polycrystalline substrates can be

3extended even qualitatively to single crystal systems Likewise there are

such sharp differences in the chemisorption behavior of hydrogen onto the

different single crystal tungsten surfaces that the specific physical process

governing adsorption on each face appears to be unique to that face This

review will therefore be concerned for the most part with literature dealing

with hydrogen adsorption on W(lOO) the gas-substrate system used in this

research

The following data review is presented in subsections each concenshy

trating on a specific experimental technique or class of experiments Subsecshy

tion 127 includes some gas mixing literature which was considered relevant

and Subsection 129 gives a bibliography of some related chemisorption

theoretical papers

122 Thermal Desorption

One technique widely employed in investigation of the kinetics of gas

456adsorption has been thermal or flash desorption By monitoring the system

5

pressure changes of a species while rapidly heating the crystal inferences

can be made on the existence of binding states adsorption energies and reaction

7 8 order Heating is accomplished by a focussed light beam electron bombardshy

7 9ment or ohmic losses bull

The strongest indication that hydrogen exhibits multiple state

adsorption behavior on W(lOO) comes from flash desorption data Tamm and

8Schmidt 7 and Madey and Yates found for room temperature adsorption that two

large peaks are observed in the plot of pressure of H2 versus temperature

These peaks are centered at 4500 K and approximately 550oK Only hydrogen

molecules are observed to be desorbed from the surface Using desorption rate

456theory the lower temperature (~l) state is found to obey first order

kinetics the peak temperature being independent of initial coverage The

higher temperature (~2) state obeys second order kinetics the peak temperature

moving down from approximately 5800 K to 5500 K as the initial coverage is

increased

In addition to the two main peaks several small peaks comprising a

total of less than 10 percent of the desorbed hydrogen are observed during a

flash when the crystal is cooled to nitrogen temperature before hydrogen

exposure These states are believed to result from adsorption onto edges and

other non-(lOO) crystal orientations 8

The dependence of the desorption energy Ed on coverage is weak

~l 8 ~2 8with Ed = 252 kcallmole and Ed = 323 kcalmole As would be predicted

from this result the ~2 state is largely populated before the ~l state begins

to fill The ratio of the populations in the states at full coverage n~1n~2

8 lOis 2111 Jelend and Menzel point out that the cross section for electron

6

desorption (see Subsection 125) from the ~1 state is considerably smaller

than that of the ~2 state a result not predicted from the experimental flash

desorption binding energies

11Tibbetts using a stepped temperature flash desorption technique

has shown that hydrogen is bound to the surface of po1ycrysta11ine tungsten

W(110) and W(lll) with a distribution of binding energies ranging from

162 kca1mo1e (07 eV) to 415 kca1mo1e (lS eV) Using this technique

desorption at a constant temperature from a distribution of states can be easily

distinguished from desorption from a state with a unique binding energy While

he did not extend his work to W(100) these results suggest that hydrogen does

not bind to the surface at a few energies but rather with a range of binding

energies Tibbetts also shows that the determination of the order of the

desorption process is difficult for adsorption into a range of binding energies

In experiments using a combination of hydrogen and deuterium as room

temperature adsorbates the elemental composition of the desorbed ~1 and ~2

states shows complete isotopic mixing between the states7S independent of the

adsorption sequence or ratios However for low temperature adsorption

complete state conversion is not observed for all the additional small peaks

Where mixing is observed atomic adsorption is suggested Where mixing is not

observed molecular adsorption is inferred

Some theoretical models indicate that multiple peak flash desorption

spectra may be due to binding energy changes caused by adsorbate-adsorbate

interactions12 or order-disorder transactions during f1ash3 13 rather than

actual distinct multiple binding states under the adsorption conditions

7

123 LEED and HEED

Low energy electron diffraction or LEED refers to the coherent

scattering of a monoenergetic focused electron beam from the surface of a

crystalline solid Electrons in the energy range 0 lt E lt 300 eV interact

strongly with the ion cores and valence electrons of the solid The inelastic

mean free path is on the order of 10 X As a consequence electrons which are

observed elastically have been backscattered within the first several surface

layers in the solid making LEED a surface sensitive probe 14 15 The periodic

translational symmetry of the surface leads to the momentum conservation laws

I

= (11)~I ~II + g

= (2mEh2)t sin reg (12)~II

where ~II is the momentum parallel to the surface of the emerging electron and

~ is the parallel momentum of the incident electron The vector g is a

reciprocal lattice vector of the solid surface and reg is the polar angle of the

incident beam As a result of the conservation laws the scattered electrons

emerge in a series of beams called a spot pattern which is a manifestation

of the surface symmetry The beams are labelled by their corresponding g

vectors Information on the actual configuration of the surface ion cores is

extracted from the plot of elastically scattered electron intensity versus

16 17incident energy called elastic intensity profiles or I-V curves

Changes in the LEED spot pattern resulting from hydrogen adsorption

18onto W(lOO) at room temperature have been reported by Estrup and Anderson and

19by Yonehara and Schmidt As hydrogen is adsorbed at room temperature the

8

(lxl) pattern characteristic of clean tungsten changes into a weak C(2x2)

pattern at a hydrogen coverage e of less than 005 monolayer (where e = 1

15 2monolayer at a coverage of 2xlO hydrogen atomsem which is 2 hydrogen atoms

for each surface tungsten atom) The (1212) spots become sharp and intense

with increased hydrogen adsorption reaching a maxima in intensity at approxishy

19 18mately 015 monolayer or e = 025 monolayer (Coverage in both cases was

determined by flash desorption) The (1212) spots split into four new spots

which move apart continuously and upon further hydrogen adsorption eventually

19form 13-order streaks at a coverage of ~ 035 monolayer These in turn

change at higher coverage into poorly developed 14-order features with the

(012) spots missing Finally near e = 10 all extra diffraction features

are gone and the LEED pattern returns to a (lxl) symmetry This series of LEED

pattern changes is reversible with coverage

For hydrogen adsorption at 1950 K the C(2x2) pattern develops

maximizes in intensity at e ~ 18 and splits into four beams similar to room

temperature adsorption However the higher coverage LEED structures third

19 0order streaks and fourth order spots are not observed At 78 K adsorption

of hydrogen produces half-order C(2x2) spots that maximize in intensity at

e ~ 027 Additional hydrogen adsorption beyond this causes the half-order

19spot intensity to decrease but with no splitting of the spots

If clean W(lOO) is allowed to cool below room temperature a sharp

19and intense C(2x2) pattern is observed even though total coverage due to the

background of H2 and CO is less than e = 001 as determined by flash desorption

measurements The intensity of these half-order beams can be altered by

additional hydrogen adsorption or by various hydrogen and oxygen heat treatmenm

9

l9Yonehara and Schmidt have postulated that the extra spots may result from

hydrogen endothermally soluble in tungsten diffusing to the surface Possible

alternative explanations include the reconstruction of the clean tungsten

surface or impurity over layer adsorption from the background It should be

20noted that Plummer and Bell saw no effects upon cooling their field emitter

tip to 78 0 K that they could attribute to hydrogen diffusion to the surface

Elastic intensity profiles for several beams have been measured by

18 21Estrup and Anderson and by Papageorgopoulos and Chen at room temperature

Estrup and Anderson reported profiles for (00) and (1212) tEED lJeams out

to 300 eV The intensity variation for the (OO) beam was found to be indepenshy

dent of hydrogen coverage a manifestation of the relatively weak scattering

cross section of a hydrogen ion core relative to that of tungsten The

additional beams observed on adsorption would be expected to be more sensitive

13to the adsorbed overlayer Estrup found the intensity of the (1212) beams

for H W(100) and D W(100) to decrease as a function of temperature between2 2

300o

K and 6000 K at a constant incident energy but neither obeyed the Debyeshy

22Waller relation Estrup interpreted this result to imply the adsorbed

species are directly involved in the diffraction process

19Yonehara examined the tEED spot intensities as a function of

temperature down to liquid nitrogen temperatures His results are examined

in the data discussion in Chapter 4

The hydrogen-WlOO) system has also been studied using high energy

electron diffraction or HEED23 This technique utilizes the strong forward

scattering of high energy electrons in probing the surface by having the

incident beam strike the solid at grazing incidence The resulting diffraction

10

patterns are sensitive to the translational symmetry of the surface As in

LEED a diffraction pattern with C(2x2) symmetry is observed during initial

hydrogen adsorption With further hydrogen exposure the beams broaden some

splitting is observed and eventually the half-order intensity decreases

completely The width of the observed half-order beams has been interpreted to

indicate hydrogen initially adsorbs into a large number of small square or

rectangular domains each consisting of 20 to 25 atoms with sides parallel to

the [100J directions

124 Electron Scattering Field Emission and Photoemission

The electronic structure of the surface layers of solids can be

characterized by electron emission spectroscopies Such techniques require a

surface sensitive perturbation on the gas-solid system leading to an electron

yield A spectrometer is used to energy analyze the emitted or scattered

24 25electrons In electron scatter~ng a monoenergetic beam of electrons is

directed onto a crystal The inelastically scattered signal containing

quantitative information about electronic excitations at the surface is

measured as a function of energy loss Field emission is a technique in which

application of a high positive electric field to the surface lowers the potenshy

26tia1 barrier for electrons to tunnel out of the system The total energy

distribution of electrons emitted as a result of the applied field contains

bull 27 28spectroscopic data on the adsorbed spec~es Structure induced by the

adsorbate in the elastic tunneling spectrum is thought to be a measure of the

d f i h f h d b 2930e1ectron~c ens~ty 0 states n t e v~c~n~ty 0 tea sor ate The

inelastic tunneling spectrum reveals discrete excitation energies in the

11

gas-surface system In photoemission or ftotoelectron spectroscopy an incident

monochromatic light beam incident on the solid induces emission of electrons by

interband transitions with or without an inelastic scattering after the initial

31 exc i tat~on

The vibrational spectrum of hydrogen adsorbed on the tungsten (100)

32surface has been observed by Propst and Piper using inelastic low energy

20electron scattering and by Plummer and Bell using field emission Propst

and Piper found two peaks in the energy loss distribution E - E pr~mary scattered

at 135 meV and 69 meV for all coverages used up to saturation These results

are interpreted as indicating hydrogen atoms are multiply bonded to tungsten

33atoms No losses near 550 meV the vibrational energy of molecular hydrogen

were observed Plummer and Bell measured the total energy distribution of

field emitted electrons from tungsten (100) as a function of coverage of

hydrogen and deuterium at several temperatures Discrete losses at 100 meV

and 50 meV for deuterium which correspond to the 135 meV and 69 meV vibrational

losses observed by Propst and Piper were evident for deuterium in the a state2

No discrete energy losses at 550 meV or 390 meV the molecular vibrational

levels of H2 and D2 respectively were measured Hence there was no observable

molecular adsorption at room temperature

Additional structure in the field emission total energy distribution

is observed as a function of hydrogen adsorption The clean surface spectrum

has two peaks 04 eV and 15 eV down from the Fermi level which have been

d ifi d h h d f d h h f 34 A h d ~ ent e as ~g ens~ty 0 states assoc~ate w~t t e sur ace s y rogen

(or deuterium) coverage increases between 0 and 5xl014 atomscm2

ie while

the ~2 state is filling the surface states disappear and the field emitted

12

energy distribution changes from that of clean tungsten to a spectrum indicative

of resonance tunneling through a level positioned 09 eV below the Fermi energy

There is a shoulder at -11 eV and an energy dependent background With

further adsorption up to saturation this energy distribution characteristic

of the ~2 state shifts to a broad level centered at 26 eV below the Fermi

level This is interpreted to indicate that the ~1 state does not adsorb on

top of the ~2 state

35The photoemission results of Wac1awski and P1ummer and Feuerbacher

36and Fitton are in agreement with those obtained using the field emission

technique As hydrogen adsorbs on the surface emission from the 04 eV

surface state decreases The level 09 eV down from the Fermi energy as well

as a level 5 eV down remain unchanged by adsorption At full coverage a strong

peak at -25 eV is observed

125 Electron Stimulated Desorption

Desorption of energetic ions and neutrals from a gas over layer can

be induced by electron bombardment This process termed electron stimulated

desorption or ESD37383940 is described below in the discussion of the data

in Chapter 4 In brief intensity and energy measurements of the ion yield can

be interpreted in terms of the parameters characterizing the binding energy

curve of the adsorbed state and adsorption states can be classified by cross

section for desorption_ ion energy distributions primary energy threshold for

desorption etc

ESD has been employed to study adsorption kinetics for the hydrogenshy

41 ~ tungsten (100) system Madey obtained a coverage dependence yield of H ions

13

observing an ion signal during the initial stages of hydrogen adsorption that

reached a maximum at a coverage of e = 017 and then decayed to zero yield

23 2after e = 075 the maximum cross section obtained was 18x10- cm at a

primary electron energy of 100 eV Tamm and Schmidt7 reported a maximum cross

section for H+ desorption of 10- 21 cm 2bull Benninghoven Loebach and Treitz42

10and Je1end and Menze1 obtained qualitatively similar (to Madey) coverage

dependencies on tungsten ribbons which were believed to be predominantly of

(100) crystallites the main difference between these results and Madeys is

the observation of a residual signal at high hydrogen coverage

The cross section results indicate that the a state even though it2

is more tightly bound than the a state provides more hydrogen ions upon1

electron impact The cross section for ion desorption from the a1 state is

d41 10-25est1mate to be 1ess t han cm2 bull

+No energy distributions of the desorbed H ions have been reported

The dependence of the desorption cross section on the primary electron energy

10has not been investigated Je1end and Menzel found their results independent

of temperature for temperatures above room temperature No ESD results for

otemperatures less than 300 K have been reported

126 Work Function

The work function is the minimum amount of energy required to remove

43 an electron from a meta1 An adsorbate may alter the potential barrier for

electrons at the surface and hence change the work function Such is the case

for hydrogen on W(100) Hydrogen adsorbs as an electronegative ion on W(100)

thereby increasing the work function

14

44Using a retarding potential current-voltage techn~que Madey and

8Yates reported a linear relation between work function change and hydrogen

coverage over the entire coverage range for adsorption occurring at a

temperature 0 f 2= 300 0 K Coverages were determined by flash desorption

Total work function change at one monolayer was found to be 090 eV which

corresponds to a constant dipole moment per adsorbed atom of 015 Debye

15 2(assuming a total coverage of 15xlO atomsem)

Estrup and Anderson18 using the retarding potential technique and

Plummer and Bel120 using a field emission Fowler-Nordheim plot45 found a

work function change versus coverage relationship which deviated only slightly

8from being linear in near agreement with the results of Madey and Yates

46 47Becker using field emission and Armstrong using a retarding potential

technique found a work function change for the H W(100) system of approximately2

48090 eV when e = 10 Hopkins and Pender obtained an anomalously low value

49of 054 eV while Collins and Blott obtained a high value of 109 eV for the

same coverage

127 Chemical Probes

Hydrogen adsorption has also been studied using a chemical probe

ie by coadsorbing some additional different gas with the hydrogen and

observing the effect this additional species had on results obtained using

techniques discussed in previous sections

50Yates and Madey found using flash desorption that CO coadsorption

onto a hydrogen monolayer displaces both the Sl and S2 binding states with

near unity probability in doing so initially replacing these states with a

15

less tightly bound hydrogen state After high CO exposure no hydrogen remains

on the surface if the coadsorption is done at room temperature At lower

temperatures only about 10 percent of the hydrogen is desorbed Jelend and

Menzel lO 5l found that the ionic and total desorption cross sections for

hydrogen the sticking probability and bond strength of hydrogen are all

affected by preadsorption of CO or oxygen Yates and Madey3 found that N2

slowly replaces preadsorbed hydrogen at temperatures greater than 300oC Below

2730

K the displacement of hydrogen by N2 does not occur

These results all show that small impurity concentrations of CO N

and 0 if intentionally introduced do affect the resulting hydrogen chemishy

sorption properties These experiments underscore the necessity of surface

chemical characterization

128 Chemisorption Models

7 18 19 52 Several models have been proposed to account for the

experimental results from hydrogen chemisorption on W(lOO) Each of these

models embodies some combination of the following three phenomenological

assumptions 1) The tungsten surface is ideal planar and free of contamishy

nants 2) There are available two types of adsorbates namely hydrogen atoms

53and molecules 3) There are three high symmetry logical bonding sites on

a BCC (100) face usually designated by the letters A B and C These

positions are A a bridge site with 2 nearest neighbors and the hydrogen

equidistant from two adjacent surface tungsten atoms B a four fold symmetric

site with 5 nearest neighbors and the hydrogen located on a surface normal that

passes through the center of a surface primitive unit cell and C the hydrogen

16

located directly above a surfac~ atom with one nearest neighbor There are

15 152x10 A sitescm2 and 1x10 B ore sites~m2 on middotW(100)

The proposed models attempt to explain or at least be consistent

with empirical results presented in Sections 121 through 127 A summary

of salient features of the reported research includes a) a linear work function

15change with coverage b) a saturation coverage of between 15 and 20x10 H

2atomscm c) a coverage dependent adsorption layer symmetry including a we11shy

ordered surface at low coverages d) the appearance at low temperatures of a

surface translational symmetry not equivalent to room temperature clean

tungsten e) adsorption into states which produces a two state thermal desorpshy

tion spectrum with relative coverages in the two states of nQ n = 21 f) a ~1 ~2

coverage dependence for the observed electronic structure including the

absence at high coverage of structure observed at low coverages and g) no

observed vibrational levels which correspond to free H2 molecule

Following is a short review of several proposed surface models

18Estrup and Anderson account for their LEED spot pattern results

using a model wherein the hydrogen molecules dissociate upon adsorption and

migrate initially to form islands of hydrogen atoms located at type A sites in

a e(2x2) structure Dissociation is assumed since the e(2x2) pattern maximizes

15 2at a coverage of 25x10 mo1ecu1escm whereas without dissociation it would

maX1m1ze at 5 1015 mo ecu1es cm2bull e site adsorptionx 1 Although type B or type

could also produce the initial e(2x2) LEED pattern these sites are rejected

15 2because of the following 1) There are only 1x10 B or e sitescm This is

not enough sites to accommodate a full monolayer coverage if all binding is

atomic and if it is into equivalent sites as suggested by the linear work

17

function change with coverage 2) While a complete 14-order LEED pattern is

possible with type B or type C sites it is not possible to produce a 14-order

54pattern with (012) spots missing using B or C sites As the adsorption

14 2proceeds past a surface density of 5xlO hydrogen atomscm the adatoms conshy

tinuously rearrange into long period superlattices with antiphase domains

that produce successively the 12-order spot splitting the 13-order streak

(only partly ordered domains) the 14-order pattern and finally the (lxl)

full coverage structure

This model has been criticized for not accounting for the two

observed binding states and for predicting a maximum in intensity of the half

order spots at e = 025 these beams maximize at less than that coverage In

addition the authors themselves questioned the idea of well-ordered super-

lattices

In order to explain the relative intensity of the two states at

saturation seen in flash desorption n~~ = 21 Tamm and Schmidt7

propose a 1 2

model wherein the initial hydrogen adsorption is atomic and at B sites in a

C(2x2) pattern At full coverage the remaining B sites are filled with

hydrogen molecules B sites are chosen over C sites using molecular orbital

55arguments A sites are rejected as there are too many to satisfy this model

This model satisfies the second order kinetics observed for the ~2 state and

the first order kinetics observed for the ~l state In addition it explains

14 2 7 their saturation coverage of gt 3xl0 moleculescm One weakness of this

model pointed out by Tamm and Schmidt is that it is not obvious the model can

account for the observed LEED and work function results Also no molecular

vibrations have been observed

18

19yonehara and Schmidt propose a three bonding state model that is

more consistent with the low temperature LEED results In their model an

additional state labeled ~3 is always present at the surface in A or B

2sites with a density of 5xl014 atomscm bull This state orders on the surface at

low temperatures to give the C(2x2) pattern observed At room and higher

temperatures it goes into solution (or perhaps disorders) becoming invisible

to LEED In the type A site model adsorption from the background fills in the

remaining A sites sequentially with the ~2 and ~l states This model is similar

18to that proposed by Estrup and Anderson except that the ~3 state although

invisible to LEED at room and higher temperatures occupies 14 of the type A

sites This model eliminates the stoichiometric objection to Estrup and

Andersons model The model depends on the possibility of type A sites being

occupied by states with considerably different properties If the ~3 state

occupies type B sites ~2 state adsorption fills in the remaining type B sites

and then ~l hydrogen adsorbs into random or possibly type C sites This

second model does not explain satisfactorily work function or LEED spot pattern

experimental data

56Propst found using a point defect type calculation that only a

B-site model of adsorption would produce the observed vibrational energies of

5769 meV and 135 meV Park using LEED intensity arguments found that only

adsorption into sites with 4-fold symmetry on the (100) surface (type B or type

C sites) would be consistent with broadening and splitting of the half-order

beams while the integral beams remain sharp

8Madey and Yates offer a critique of several of the proposed models

found No evidence was found from work function and electron scattering

experiments to support the hypothesis that the low temperature c(2x2)

pattern results from hydrogen on the surface In addition the fast hydrogen

state is not believed to be responsible for the c(2x2) pattern LEED

intensity profiles from the undosed cold surface c(2x2) pattern and from the

c(2x2) pattern associated with the ~2 hydrogen state are not identical an

indication that the atomic geometry responsible for the two patterns may not

be the same

Electron scattering energy loss spectra have been obtained as a

function of hydrogen coverage for loss energies of 0 eV ~ w lt 40 eV Hydrogen

adsorption has a marked effect on the spectra Fine structure in the data

has been discussed and compared to photoemission data Cooling the clean

crystal below room temperature causes only a slight change in the energy loss

spectrum

iii

ACKNOWLEDGMENTS

I would like to sincerely thank my advisor Professor F M Propst

for guidance given throughout the course of this research and for his knowledge

of surface research which helped shape this work Dr C B Duke for many

helpful conversations and for assistance in the analysis of the data my wife

Lois for her patience and steadfast encouragement my parents and parents-inshy

law for their generous and needed support Paul Luscher and John Wendelken for

their creativity and cooperation in the design of the apparatus and for many

interesting scientific discussions Jim Burkstrand and Chris Foster for their

friendship and technical help on many occasions Bob Bales George Bouck Bob

Fults Bill Beaulin Lyle Bandy and Bill Lawrence for design assistance and

expert workmanship in the construction of the apparatus Nick Vassos and Bill

Thrasher for the imaginative and superior work in materials processing necesshy

sary in the construction of the instrument Gene Gardner and co-workers for

able assistance in electronics Bob MacFarlane and assistants for their high

quality draftmanship Jack Gladin and Jim Gray for an excellent job on the

photography needed for this research Bob Ebeling for his cooperation in the

print shop Chris Fleener for an excellent typing of the manuscript and the

rest of the staff of the Coordinated Science Laboratory whose efforts have

been gratefully appreciated during my tenure as a graduate student

bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

iv

TABLE OF CONTENTS

Chapter Page

1 MOTIVATION AND REVIEW OF LITERATURE bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 1

11 Introduction and Motivationbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 1

12 Review of Literature bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 4o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

121 Introductionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 4 122 Thermal Desorptionbull o 4 123 LEED and HEEDbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 7 124 Electron Scattering Field Emission and

Photoemissionbullbullbull 100 bullbullbullbullbull 0 bullbullbullbull

125 Electron Stimulated Desorptionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 12 126 Work Functionbullbullbullbullbullbullbullbullbull 13o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

1 27 Chemical Probes bullbullbullbullbullbullbull bullbullbullbull bullbullbullbull 14I) O

128 Chemisorption Models bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 15 1 29 Theoretical Treatment bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 19 1 210 Literature Surnmaryo 20

13 Review of Experimental Techniques bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 21

2 EXPERIMENTAL APPARATUS bullbullbullbullbullbullbullbullbullbullbullbull ~ bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 22

21 CGeneral Description bull 22

22 Vacuum System and Magnetic Shieldingbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 22

23 Electron Gun and Gun Electronics bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 25

24 Energy Spectrometer 29

25 Spectrometer Electronics bullbull 45o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

26 LEED Display System 490 bullbullbull 0 bullbullbullbullbullbullbull 0 bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull I)

27 Crystal Manipulator Assembly bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 51

28 Crystal Preparation and Cleaning Procedures bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 57

29 Gas Handling System o bullbullbullbullbullbullbullbullbullbullbullbullbull o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 62

210 Mass Spectrometer bullbullbullbull o 64

3 EXPERIMENTAL PROCEDURE AND RESULTS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 65

31 Introductionbullbullbullbullbullbullbullbullbullbullbullbull 65

32 Experimental Procedure o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull o bullbullbullbullbullbullbull 65 321 Work Function and Coverage Measurements bullbullbullbullbullbullbullbullbullbullbullbullbull 65 322 Primary Electron Beam Energy and Current

Measurements 6811 bullbullbullbullbullbullbullbullbullbullbullbullbullbullbull bullbullbullbull

323 Target Temperature Measurement bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 69 324 Electron Scattering and Auger Measurements bullbullbullbullbullbullbullbullbullbull 69 325 Ion Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 72o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

v

33 Ion Species Identificationbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 74

34 Work Function Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 77

35 u+ Ion Desorption Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 83 351 ut Ion Energy Distributionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 83 352 Desorption Curve Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 93 353 Temperature Dependence of the Fast State a+ Ion

Signal 4) 96

354 Additional Measurements on the Fast State bullbullbullbullbullbullbullbullbullbull 98

36 Low Energy Electron Diffraction Data bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull lll

37 Electron Scattering Data bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 119

4 DISCUSSION OF RESULTS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 126

41 Auger Electron Spectroscopy Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 126

42 Work Function Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 127

43 Electron Stimulated Desorption Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 129 431 Room Temperature Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 130 432 Low Temperature Fast State Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 135 433 Low Temperature Dosed Surface Results bullbullbullbullbullbullbullbullbullbullbullbullbull142

44 Low Energy Electron Diffraction Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 145

45 Electron Scattering Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 161

5 SUMMARY AND CONCLUSIONS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull165

REFERENCES 171

APPENDIX A 0 176 poundI

APPENDIX B 9 c 179

APPENDIX C 0 182 0 bull

VITA 190lilt bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull bullbull

1

CHAPTER 1

MOTIVATION AND REVIEW OF LITERATURE

11 Introduction and Motivation

The experimental investigation of chemisorption of gases onto metallic

surfaces has uncovered in the past decade a baffling complexity of adsorption

geometries and binding state kinetics and energetics A growing interest in

surface phenomena generated by the questions raised in these researches and

by the long anticipated use of the answers has led to the rapid development

of new surface probing techniques and to an increase in the availability of new

technologies compatible with surface research Aided by the new technologies

and techniques chemisorption research has been undertaken on many different

systems (system =solid substrate + gas adsorbate) under a wide range of

environmental conditions Surface theoreticians have proposed microscopic and

phenomenological models with which data have been analyzed and compared A

large experimental and theoretical literature in the field has resulted

The study of adsorption on metal surfaces has among its primary

objectives a detailed description of the physical geometry of the atoms in the

surface layers and a quantitative understanding of electronic and chemical

surface phenomena While considerable progress has been made in documenting

gas-surface phenomena it has been only in the past three years that any

credible claims to understanding specific surface overlayer geometries have

1been made At the same time many basic questions involving system kinetics

and energetics have not been adequately resolved 2

It has become apparent from the failure of the research effort to

satisfy the above objectives and from seemingly conflicting data that adequate

2

characterization is basic to all surface experimentation The need for such

characterization has been strongly emphasized by disagreement between quasishy

similar research performed at different locations Small impurity concentrashy

tions small changes in crystal orientation minor temperature differences etc

often have considerable effects on experimental measurements

A corollary to the necessity of being able to characterize the

surface is the importance of being able to control the surface characterization

In order to compare data it is necessary to reproduce all experimental paramshy

eters accurately The theory to which experimental data are compared is

constrained by complexity to be based on ideal clean substrates with a

definable gas-surface interaction It is desirable to be able to choose the

adsorbate-substrate system and experimental parameters that can also be used

with theoretical models Of course compromises are inevitable

One approach in surface research has been to incorporate several

experimental techniques into one apparatus At the present time this approach

is increasingly more common A combination of surface sensitive probes offers

many advantages It provides for more checks of the data against results

published in the literature It allows for better characterization of the

experimental parameters It allows surface phenomena to be probed in more than

one way in order to eliminate ambiguities andor to reinforce conclusions It

eliminates the uncertainties that arise from data taken at different locations

using the techniques separately hence it allows better correlations between

data taken using the different techniques

In this study several aspects of the chemisorption interaction of

hydrogen with the low index (100) plane of single crystal tungsten denoted

3

W(100) have been investigated The choice of the tungsten-hydrogen system is

motivated by several factors The literature raises questions on the unusual

state behavior observed when hydrogen adsorbs on W(100) questions which our

apparatus is especially well-suited to study In addition the interaction of

hydrogen with close packed planes of single crystal metals including W(100)

is of particular interest to theoreticians since hydrogen is the simplest

molecular gas as detailed in Sections 128 and 129 below recent theories

of chemisorption have been applied to the hydrogen-tungsten system Tungsten

was chosen as the substrate metal because of its relative ease of preparation

and cleaning and because the large body of literature on the hydrogen-tungsten

system indicates wide interest in this interaction

In particular this research is concentrated in three areas Tung~

sten when cleaned and subsequently cooled to below room temperature exhibits

an anomalous centered (2x2) denoted C(2x2) low energy electron diffraction

pattern The possibility that this is caused by hydrogen diffusing from the

bulk or adsorbing from the background has been investigated Secondly the

room temperature and low temperature hydrogen adsorption binding states have

been studied using electron stimulated desorption and electron spectroscopy

Finally low energy electron diffraction has been used in an investigation of

the atomic geometry of the clean tungsten surface and the geometry of one of

the hydrogen binding states

The remaining sections of this chapter summarize literature relating

to this research Chapter 2 details the construction and performance of the

experimental apparatus A presentation of the experimental data is made in

Chapter 3 A discussion of the results is made in Chapter 4 A summaryof

4

the research and conclusions from it are given in Chapter 5

12 Review of Literature

121 Introduction

Experimental investigation of the adsorption of hydrogen on tungsten

has been carried out using field emitter tips macroscopic polycrystalline

surfaces and all the low index orientation single crystal faces including

W(lOO) From the wide range of results it is clear that the substrate exerts

a strong influence on the adsorption properties The question has been raised

whether adsorption models postulated from polycrystalline substrates can be

3extended even qualitatively to single crystal systems Likewise there are

such sharp differences in the chemisorption behavior of hydrogen onto the

different single crystal tungsten surfaces that the specific physical process

governing adsorption on each face appears to be unique to that face This

review will therefore be concerned for the most part with literature dealing

with hydrogen adsorption on W(lOO) the gas-substrate system used in this

research

The following data review is presented in subsections each concenshy

trating on a specific experimental technique or class of experiments Subsecshy

tion 127 includes some gas mixing literature which was considered relevant

and Subsection 129 gives a bibliography of some related chemisorption

theoretical papers

122 Thermal Desorption

One technique widely employed in investigation of the kinetics of gas

456adsorption has been thermal or flash desorption By monitoring the system

5

pressure changes of a species while rapidly heating the crystal inferences

can be made on the existence of binding states adsorption energies and reaction

7 8 order Heating is accomplished by a focussed light beam electron bombardshy

7 9ment or ohmic losses bull

The strongest indication that hydrogen exhibits multiple state

adsorption behavior on W(lOO) comes from flash desorption data Tamm and

8Schmidt 7 and Madey and Yates found for room temperature adsorption that two

large peaks are observed in the plot of pressure of H2 versus temperature

These peaks are centered at 4500 K and approximately 550oK Only hydrogen

molecules are observed to be desorbed from the surface Using desorption rate

456theory the lower temperature (~l) state is found to obey first order

kinetics the peak temperature being independent of initial coverage The

higher temperature (~2) state obeys second order kinetics the peak temperature

moving down from approximately 5800 K to 5500 K as the initial coverage is

increased

In addition to the two main peaks several small peaks comprising a

total of less than 10 percent of the desorbed hydrogen are observed during a

flash when the crystal is cooled to nitrogen temperature before hydrogen

exposure These states are believed to result from adsorption onto edges and

other non-(lOO) crystal orientations 8

The dependence of the desorption energy Ed on coverage is weak

~l 8 ~2 8with Ed = 252 kcallmole and Ed = 323 kcalmole As would be predicted

from this result the ~2 state is largely populated before the ~l state begins

to fill The ratio of the populations in the states at full coverage n~1n~2

8 lOis 2111 Jelend and Menzel point out that the cross section for electron

6

desorption (see Subsection 125) from the ~1 state is considerably smaller

than that of the ~2 state a result not predicted from the experimental flash

desorption binding energies

11Tibbetts using a stepped temperature flash desorption technique

has shown that hydrogen is bound to the surface of po1ycrysta11ine tungsten

W(110) and W(lll) with a distribution of binding energies ranging from

162 kca1mo1e (07 eV) to 415 kca1mo1e (lS eV) Using this technique

desorption at a constant temperature from a distribution of states can be easily

distinguished from desorption from a state with a unique binding energy While

he did not extend his work to W(100) these results suggest that hydrogen does

not bind to the surface at a few energies but rather with a range of binding

energies Tibbetts also shows that the determination of the order of the

desorption process is difficult for adsorption into a range of binding energies

In experiments using a combination of hydrogen and deuterium as room

temperature adsorbates the elemental composition of the desorbed ~1 and ~2

states shows complete isotopic mixing between the states7S independent of the

adsorption sequence or ratios However for low temperature adsorption

complete state conversion is not observed for all the additional small peaks

Where mixing is observed atomic adsorption is suggested Where mixing is not

observed molecular adsorption is inferred

Some theoretical models indicate that multiple peak flash desorption

spectra may be due to binding energy changes caused by adsorbate-adsorbate

interactions12 or order-disorder transactions during f1ash3 13 rather than

actual distinct multiple binding states under the adsorption conditions

7

123 LEED and HEED

Low energy electron diffraction or LEED refers to the coherent

scattering of a monoenergetic focused electron beam from the surface of a

crystalline solid Electrons in the energy range 0 lt E lt 300 eV interact

strongly with the ion cores and valence electrons of the solid The inelastic

mean free path is on the order of 10 X As a consequence electrons which are

observed elastically have been backscattered within the first several surface

layers in the solid making LEED a surface sensitive probe 14 15 The periodic

translational symmetry of the surface leads to the momentum conservation laws

I

= (11)~I ~II + g

= (2mEh2)t sin reg (12)~II

where ~II is the momentum parallel to the surface of the emerging electron and

~ is the parallel momentum of the incident electron The vector g is a

reciprocal lattice vector of the solid surface and reg is the polar angle of the

incident beam As a result of the conservation laws the scattered electrons

emerge in a series of beams called a spot pattern which is a manifestation

of the surface symmetry The beams are labelled by their corresponding g

vectors Information on the actual configuration of the surface ion cores is

extracted from the plot of elastically scattered electron intensity versus

16 17incident energy called elastic intensity profiles or I-V curves

Changes in the LEED spot pattern resulting from hydrogen adsorption

18onto W(lOO) at room temperature have been reported by Estrup and Anderson and

19by Yonehara and Schmidt As hydrogen is adsorbed at room temperature the

8

(lxl) pattern characteristic of clean tungsten changes into a weak C(2x2)

pattern at a hydrogen coverage e of less than 005 monolayer (where e = 1

15 2monolayer at a coverage of 2xlO hydrogen atomsem which is 2 hydrogen atoms

for each surface tungsten atom) The (1212) spots become sharp and intense

with increased hydrogen adsorption reaching a maxima in intensity at approxishy

19 18mately 015 monolayer or e = 025 monolayer (Coverage in both cases was

determined by flash desorption) The (1212) spots split into four new spots

which move apart continuously and upon further hydrogen adsorption eventually

19form 13-order streaks at a coverage of ~ 035 monolayer These in turn

change at higher coverage into poorly developed 14-order features with the

(012) spots missing Finally near e = 10 all extra diffraction features

are gone and the LEED pattern returns to a (lxl) symmetry This series of LEED

pattern changes is reversible with coverage

For hydrogen adsorption at 1950 K the C(2x2) pattern develops

maximizes in intensity at e ~ 18 and splits into four beams similar to room

temperature adsorption However the higher coverage LEED structures third

19 0order streaks and fourth order spots are not observed At 78 K adsorption

of hydrogen produces half-order C(2x2) spots that maximize in intensity at

e ~ 027 Additional hydrogen adsorption beyond this causes the half-order

19spot intensity to decrease but with no splitting of the spots

If clean W(lOO) is allowed to cool below room temperature a sharp

19and intense C(2x2) pattern is observed even though total coverage due to the

background of H2 and CO is less than e = 001 as determined by flash desorption

measurements The intensity of these half-order beams can be altered by

additional hydrogen adsorption or by various hydrogen and oxygen heat treatmenm

9

l9Yonehara and Schmidt have postulated that the extra spots may result from

hydrogen endothermally soluble in tungsten diffusing to the surface Possible

alternative explanations include the reconstruction of the clean tungsten

surface or impurity over layer adsorption from the background It should be

20noted that Plummer and Bell saw no effects upon cooling their field emitter

tip to 78 0 K that they could attribute to hydrogen diffusion to the surface

Elastic intensity profiles for several beams have been measured by

18 21Estrup and Anderson and by Papageorgopoulos and Chen at room temperature

Estrup and Anderson reported profiles for (00) and (1212) tEED lJeams out

to 300 eV The intensity variation for the (OO) beam was found to be indepenshy

dent of hydrogen coverage a manifestation of the relatively weak scattering

cross section of a hydrogen ion core relative to that of tungsten The

additional beams observed on adsorption would be expected to be more sensitive

13to the adsorbed overlayer Estrup found the intensity of the (1212) beams

for H W(100) and D W(100) to decrease as a function of temperature between2 2

300o

K and 6000 K at a constant incident energy but neither obeyed the Debyeshy

22Waller relation Estrup interpreted this result to imply the adsorbed

species are directly involved in the diffraction process

19Yonehara examined the tEED spot intensities as a function of

temperature down to liquid nitrogen temperatures His results are examined

in the data discussion in Chapter 4

The hydrogen-WlOO) system has also been studied using high energy

electron diffraction or HEED23 This technique utilizes the strong forward

scattering of high energy electrons in probing the surface by having the

incident beam strike the solid at grazing incidence The resulting diffraction

10

patterns are sensitive to the translational symmetry of the surface As in

LEED a diffraction pattern with C(2x2) symmetry is observed during initial

hydrogen adsorption With further hydrogen exposure the beams broaden some

splitting is observed and eventually the half-order intensity decreases

completely The width of the observed half-order beams has been interpreted to

indicate hydrogen initially adsorbs into a large number of small square or

rectangular domains each consisting of 20 to 25 atoms with sides parallel to

the [100J directions

124 Electron Scattering Field Emission and Photoemission

The electronic structure of the surface layers of solids can be

characterized by electron emission spectroscopies Such techniques require a

surface sensitive perturbation on the gas-solid system leading to an electron

yield A spectrometer is used to energy analyze the emitted or scattered

24 25electrons In electron scatter~ng a monoenergetic beam of electrons is

directed onto a crystal The inelastically scattered signal containing

quantitative information about electronic excitations at the surface is

measured as a function of energy loss Field emission is a technique in which

application of a high positive electric field to the surface lowers the potenshy

26tia1 barrier for electrons to tunnel out of the system The total energy

distribution of electrons emitted as a result of the applied field contains

bull 27 28spectroscopic data on the adsorbed spec~es Structure induced by the

adsorbate in the elastic tunneling spectrum is thought to be a measure of the

d f i h f h d b 2930e1ectron~c ens~ty 0 states n t e v~c~n~ty 0 tea sor ate The

inelastic tunneling spectrum reveals discrete excitation energies in the

11

gas-surface system In photoemission or ftotoelectron spectroscopy an incident

monochromatic light beam incident on the solid induces emission of electrons by

interband transitions with or without an inelastic scattering after the initial

31 exc i tat~on

The vibrational spectrum of hydrogen adsorbed on the tungsten (100)

32surface has been observed by Propst and Piper using inelastic low energy

20electron scattering and by Plummer and Bell using field emission Propst

and Piper found two peaks in the energy loss distribution E - E pr~mary scattered

at 135 meV and 69 meV for all coverages used up to saturation These results

are interpreted as indicating hydrogen atoms are multiply bonded to tungsten

33atoms No losses near 550 meV the vibrational energy of molecular hydrogen

were observed Plummer and Bell measured the total energy distribution of

field emitted electrons from tungsten (100) as a function of coverage of

hydrogen and deuterium at several temperatures Discrete losses at 100 meV

and 50 meV for deuterium which correspond to the 135 meV and 69 meV vibrational

losses observed by Propst and Piper were evident for deuterium in the a state2

No discrete energy losses at 550 meV or 390 meV the molecular vibrational

levels of H2 and D2 respectively were measured Hence there was no observable

molecular adsorption at room temperature

Additional structure in the field emission total energy distribution

is observed as a function of hydrogen adsorption The clean surface spectrum

has two peaks 04 eV and 15 eV down from the Fermi level which have been

d ifi d h h d f d h h f 34 A h d ~ ent e as ~g ens~ty 0 states assoc~ate w~t t e sur ace s y rogen

(or deuterium) coverage increases between 0 and 5xl014 atomscm2

ie while

the ~2 state is filling the surface states disappear and the field emitted

12

energy distribution changes from that of clean tungsten to a spectrum indicative

of resonance tunneling through a level positioned 09 eV below the Fermi energy

There is a shoulder at -11 eV and an energy dependent background With

further adsorption up to saturation this energy distribution characteristic

of the ~2 state shifts to a broad level centered at 26 eV below the Fermi

level This is interpreted to indicate that the ~1 state does not adsorb on

top of the ~2 state

35The photoemission results of Wac1awski and P1ummer and Feuerbacher

36and Fitton are in agreement with those obtained using the field emission

technique As hydrogen adsorbs on the surface emission from the 04 eV

surface state decreases The level 09 eV down from the Fermi energy as well

as a level 5 eV down remain unchanged by adsorption At full coverage a strong

peak at -25 eV is observed

125 Electron Stimulated Desorption

Desorption of energetic ions and neutrals from a gas over layer can

be induced by electron bombardment This process termed electron stimulated

desorption or ESD37383940 is described below in the discussion of the data

in Chapter 4 In brief intensity and energy measurements of the ion yield can

be interpreted in terms of the parameters characterizing the binding energy

curve of the adsorbed state and adsorption states can be classified by cross

section for desorption_ ion energy distributions primary energy threshold for

desorption etc

ESD has been employed to study adsorption kinetics for the hydrogenshy

41 ~ tungsten (100) system Madey obtained a coverage dependence yield of H ions

13

observing an ion signal during the initial stages of hydrogen adsorption that

reached a maximum at a coverage of e = 017 and then decayed to zero yield

23 2after e = 075 the maximum cross section obtained was 18x10- cm at a

primary electron energy of 100 eV Tamm and Schmidt7 reported a maximum cross

section for H+ desorption of 10- 21 cm 2bull Benninghoven Loebach and Treitz42

10and Je1end and Menze1 obtained qualitatively similar (to Madey) coverage

dependencies on tungsten ribbons which were believed to be predominantly of

(100) crystallites the main difference between these results and Madeys is

the observation of a residual signal at high hydrogen coverage

The cross section results indicate that the a state even though it2

is more tightly bound than the a state provides more hydrogen ions upon1

electron impact The cross section for ion desorption from the a1 state is

d41 10-25est1mate to be 1ess t han cm2 bull

+No energy distributions of the desorbed H ions have been reported

The dependence of the desorption cross section on the primary electron energy

10has not been investigated Je1end and Menzel found their results independent

of temperature for temperatures above room temperature No ESD results for

otemperatures less than 300 K have been reported

126 Work Function

The work function is the minimum amount of energy required to remove

43 an electron from a meta1 An adsorbate may alter the potential barrier for

electrons at the surface and hence change the work function Such is the case

for hydrogen on W(100) Hydrogen adsorbs as an electronegative ion on W(100)

thereby increasing the work function

14

44Using a retarding potential current-voltage techn~que Madey and

8Yates reported a linear relation between work function change and hydrogen

coverage over the entire coverage range for adsorption occurring at a

temperature 0 f 2= 300 0 K Coverages were determined by flash desorption

Total work function change at one monolayer was found to be 090 eV which

corresponds to a constant dipole moment per adsorbed atom of 015 Debye

15 2(assuming a total coverage of 15xlO atomsem)

Estrup and Anderson18 using the retarding potential technique and

Plummer and Bel120 using a field emission Fowler-Nordheim plot45 found a

work function change versus coverage relationship which deviated only slightly

8from being linear in near agreement with the results of Madey and Yates

46 47Becker using field emission and Armstrong using a retarding potential

technique found a work function change for the H W(100) system of approximately2

48090 eV when e = 10 Hopkins and Pender obtained an anomalously low value

49of 054 eV while Collins and Blott obtained a high value of 109 eV for the

same coverage

127 Chemical Probes

Hydrogen adsorption has also been studied using a chemical probe

ie by coadsorbing some additional different gas with the hydrogen and

observing the effect this additional species had on results obtained using

techniques discussed in previous sections

50Yates and Madey found using flash desorption that CO coadsorption

onto a hydrogen monolayer displaces both the Sl and S2 binding states with

near unity probability in doing so initially replacing these states with a

15

less tightly bound hydrogen state After high CO exposure no hydrogen remains

on the surface if the coadsorption is done at room temperature At lower

temperatures only about 10 percent of the hydrogen is desorbed Jelend and

Menzel lO 5l found that the ionic and total desorption cross sections for

hydrogen the sticking probability and bond strength of hydrogen are all

affected by preadsorption of CO or oxygen Yates and Madey3 found that N2

slowly replaces preadsorbed hydrogen at temperatures greater than 300oC Below

2730

K the displacement of hydrogen by N2 does not occur

These results all show that small impurity concentrations of CO N

and 0 if intentionally introduced do affect the resulting hydrogen chemishy

sorption properties These experiments underscore the necessity of surface

chemical characterization

128 Chemisorption Models

7 18 19 52 Several models have been proposed to account for the

experimental results from hydrogen chemisorption on W(lOO) Each of these

models embodies some combination of the following three phenomenological

assumptions 1) The tungsten surface is ideal planar and free of contamishy

nants 2) There are available two types of adsorbates namely hydrogen atoms

53and molecules 3) There are three high symmetry logical bonding sites on

a BCC (100) face usually designated by the letters A B and C These

positions are A a bridge site with 2 nearest neighbors and the hydrogen

equidistant from two adjacent surface tungsten atoms B a four fold symmetric

site with 5 nearest neighbors and the hydrogen located on a surface normal that

passes through the center of a surface primitive unit cell and C the hydrogen

16

located directly above a surfac~ atom with one nearest neighbor There are

15 152x10 A sitescm2 and 1x10 B ore sites~m2 on middotW(100)

The proposed models attempt to explain or at least be consistent

with empirical results presented in Sections 121 through 127 A summary

of salient features of the reported research includes a) a linear work function

15change with coverage b) a saturation coverage of between 15 and 20x10 H

2atomscm c) a coverage dependent adsorption layer symmetry including a we11shy

ordered surface at low coverages d) the appearance at low temperatures of a

surface translational symmetry not equivalent to room temperature clean

tungsten e) adsorption into states which produces a two state thermal desorpshy

tion spectrum with relative coverages in the two states of nQ n = 21 f) a ~1 ~2

coverage dependence for the observed electronic structure including the

absence at high coverage of structure observed at low coverages and g) no

observed vibrational levels which correspond to free H2 molecule

Following is a short review of several proposed surface models

18Estrup and Anderson account for their LEED spot pattern results

using a model wherein the hydrogen molecules dissociate upon adsorption and

migrate initially to form islands of hydrogen atoms located at type A sites in

a e(2x2) structure Dissociation is assumed since the e(2x2) pattern maximizes

15 2at a coverage of 25x10 mo1ecu1escm whereas without dissociation it would

maX1m1ze at 5 1015 mo ecu1es cm2bull e site adsorptionx 1 Although type B or type

could also produce the initial e(2x2) LEED pattern these sites are rejected

15 2because of the following 1) There are only 1x10 B or e sitescm This is

not enough sites to accommodate a full monolayer coverage if all binding is

atomic and if it is into equivalent sites as suggested by the linear work

17

function change with coverage 2) While a complete 14-order LEED pattern is

possible with type B or type C sites it is not possible to produce a 14-order

54pattern with (012) spots missing using B or C sites As the adsorption

14 2proceeds past a surface density of 5xlO hydrogen atomscm the adatoms conshy

tinuously rearrange into long period superlattices with antiphase domains

that produce successively the 12-order spot splitting the 13-order streak

(only partly ordered domains) the 14-order pattern and finally the (lxl)

full coverage structure

This model has been criticized for not accounting for the two

observed binding states and for predicting a maximum in intensity of the half

order spots at e = 025 these beams maximize at less than that coverage In

addition the authors themselves questioned the idea of well-ordered super-

lattices

In order to explain the relative intensity of the two states at

saturation seen in flash desorption n~~ = 21 Tamm and Schmidt7

propose a 1 2

model wherein the initial hydrogen adsorption is atomic and at B sites in a

C(2x2) pattern At full coverage the remaining B sites are filled with

hydrogen molecules B sites are chosen over C sites using molecular orbital

55arguments A sites are rejected as there are too many to satisfy this model

This model satisfies the second order kinetics observed for the ~2 state and

the first order kinetics observed for the ~l state In addition it explains

14 2 7 their saturation coverage of gt 3xl0 moleculescm One weakness of this

model pointed out by Tamm and Schmidt is that it is not obvious the model can

account for the observed LEED and work function results Also no molecular

vibrations have been observed

18

19yonehara and Schmidt propose a three bonding state model that is

more consistent with the low temperature LEED results In their model an

additional state labeled ~3 is always present at the surface in A or B

2sites with a density of 5xl014 atomscm bull This state orders on the surface at

low temperatures to give the C(2x2) pattern observed At room and higher

temperatures it goes into solution (or perhaps disorders) becoming invisible

to LEED In the type A site model adsorption from the background fills in the

remaining A sites sequentially with the ~2 and ~l states This model is similar

18to that proposed by Estrup and Anderson except that the ~3 state although

invisible to LEED at room and higher temperatures occupies 14 of the type A

sites This model eliminates the stoichiometric objection to Estrup and

Andersons model The model depends on the possibility of type A sites being

occupied by states with considerably different properties If the ~3 state

occupies type B sites ~2 state adsorption fills in the remaining type B sites

and then ~l hydrogen adsorbs into random or possibly type C sites This

second model does not explain satisfactorily work function or LEED spot pattern

experimental data

56Propst found using a point defect type calculation that only a

B-site model of adsorption would produce the observed vibrational energies of

5769 meV and 135 meV Park using LEED intensity arguments found that only

adsorption into sites with 4-fold symmetry on the (100) surface (type B or type

C sites) would be consistent with broadening and splitting of the half-order

beams while the integral beams remain sharp

8Madey and Yates offer a critique of several of the proposed models

iii

ACKNOWLEDGMENTS

I would like to sincerely thank my advisor Professor F M Propst

for guidance given throughout the course of this research and for his knowledge

of surface research which helped shape this work Dr C B Duke for many

helpful conversations and for assistance in the analysis of the data my wife

Lois for her patience and steadfast encouragement my parents and parents-inshy

law for their generous and needed support Paul Luscher and John Wendelken for

their creativity and cooperation in the design of the apparatus and for many

interesting scientific discussions Jim Burkstrand and Chris Foster for their

friendship and technical help on many occasions Bob Bales George Bouck Bob

Fults Bill Beaulin Lyle Bandy and Bill Lawrence for design assistance and

expert workmanship in the construction of the apparatus Nick Vassos and Bill

Thrasher for the imaginative and superior work in materials processing necesshy

sary in the construction of the instrument Gene Gardner and co-workers for

able assistance in electronics Bob MacFarlane and assistants for their high

quality draftmanship Jack Gladin and Jim Gray for an excellent job on the

photography needed for this research Bob Ebeling for his cooperation in the

print shop Chris Fleener for an excellent typing of the manuscript and the

rest of the staff of the Coordinated Science Laboratory whose efforts have

been gratefully appreciated during my tenure as a graduate student

bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

iv

TABLE OF CONTENTS

Chapter Page

1 MOTIVATION AND REVIEW OF LITERATURE bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 1

11 Introduction and Motivationbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 1

12 Review of Literature bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 4o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

121 Introductionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 4 122 Thermal Desorptionbull o 4 123 LEED and HEEDbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 7 124 Electron Scattering Field Emission and

Photoemissionbullbullbull 100 bullbullbullbullbull 0 bullbullbullbull

125 Electron Stimulated Desorptionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 12 126 Work Functionbullbullbullbullbullbullbullbullbull 13o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

1 27 Chemical Probes bullbullbullbullbullbullbull bullbullbullbull bullbullbullbull 14I) O

128 Chemisorption Models bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 15 1 29 Theoretical Treatment bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 19 1 210 Literature Surnmaryo 20

13 Review of Experimental Techniques bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 21

2 EXPERIMENTAL APPARATUS bullbullbullbullbullbullbullbullbullbullbullbull ~ bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 22

21 CGeneral Description bull 22

22 Vacuum System and Magnetic Shieldingbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 22

23 Electron Gun and Gun Electronics bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 25

24 Energy Spectrometer 29

25 Spectrometer Electronics bullbull 45o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

26 LEED Display System 490 bullbullbull 0 bullbullbullbullbullbullbull 0 bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull I)

27 Crystal Manipulator Assembly bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 51

28 Crystal Preparation and Cleaning Procedures bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 57

29 Gas Handling System o bullbullbullbullbullbullbullbullbullbullbullbullbull o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 62

210 Mass Spectrometer bullbullbullbull o 64

3 EXPERIMENTAL PROCEDURE AND RESULTS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 65

31 Introductionbullbullbullbullbullbullbullbullbullbullbullbull 65

32 Experimental Procedure o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull o bullbullbullbullbullbullbull 65 321 Work Function and Coverage Measurements bullbullbullbullbullbullbullbullbullbullbullbullbull 65 322 Primary Electron Beam Energy and Current

Measurements 6811 bullbullbullbullbullbullbullbullbullbullbullbullbullbullbull bullbullbullbull

323 Target Temperature Measurement bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 69 324 Electron Scattering and Auger Measurements bullbullbullbullbullbullbullbullbullbull 69 325 Ion Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 72o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

v

33 Ion Species Identificationbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 74

34 Work Function Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 77

35 u+ Ion Desorption Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 83 351 ut Ion Energy Distributionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 83 352 Desorption Curve Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 93 353 Temperature Dependence of the Fast State a+ Ion

Signal 4) 96

354 Additional Measurements on the Fast State bullbullbullbullbullbullbullbullbullbull 98

36 Low Energy Electron Diffraction Data bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull lll

37 Electron Scattering Data bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 119

4 DISCUSSION OF RESULTS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 126

41 Auger Electron Spectroscopy Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 126

42 Work Function Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 127

43 Electron Stimulated Desorption Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 129 431 Room Temperature Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 130 432 Low Temperature Fast State Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 135 433 Low Temperature Dosed Surface Results bullbullbullbullbullbullbullbullbullbullbullbullbull142

44 Low Energy Electron Diffraction Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 145

45 Electron Scattering Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 161

5 SUMMARY AND CONCLUSIONS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull165

REFERENCES 171

APPENDIX A 0 176 poundI

APPENDIX B 9 c 179

APPENDIX C 0 182 0 bull

VITA 190lilt bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull bullbull

1

CHAPTER 1

MOTIVATION AND REVIEW OF LITERATURE

11 Introduction and Motivation

The experimental investigation of chemisorption of gases onto metallic

surfaces has uncovered in the past decade a baffling complexity of adsorption

geometries and binding state kinetics and energetics A growing interest in

surface phenomena generated by the questions raised in these researches and

by the long anticipated use of the answers has led to the rapid development

of new surface probing techniques and to an increase in the availability of new

technologies compatible with surface research Aided by the new technologies

and techniques chemisorption research has been undertaken on many different

systems (system =solid substrate + gas adsorbate) under a wide range of

environmental conditions Surface theoreticians have proposed microscopic and

phenomenological models with which data have been analyzed and compared A

large experimental and theoretical literature in the field has resulted

The study of adsorption on metal surfaces has among its primary

objectives a detailed description of the physical geometry of the atoms in the

surface layers and a quantitative understanding of electronic and chemical

surface phenomena While considerable progress has been made in documenting

gas-surface phenomena it has been only in the past three years that any

credible claims to understanding specific surface overlayer geometries have

1been made At the same time many basic questions involving system kinetics

and energetics have not been adequately resolved 2

It has become apparent from the failure of the research effort to

satisfy the above objectives and from seemingly conflicting data that adequate

2

characterization is basic to all surface experimentation The need for such

characterization has been strongly emphasized by disagreement between quasishy

similar research performed at different locations Small impurity concentrashy

tions small changes in crystal orientation minor temperature differences etc

often have considerable effects on experimental measurements

A corollary to the necessity of being able to characterize the

surface is the importance of being able to control the surface characterization

In order to compare data it is necessary to reproduce all experimental paramshy

eters accurately The theory to which experimental data are compared is

constrained by complexity to be based on ideal clean substrates with a

definable gas-surface interaction It is desirable to be able to choose the

adsorbate-substrate system and experimental parameters that can also be used

with theoretical models Of course compromises are inevitable

One approach in surface research has been to incorporate several

experimental techniques into one apparatus At the present time this approach

is increasingly more common A combination of surface sensitive probes offers

many advantages It provides for more checks of the data against results

published in the literature It allows for better characterization of the

experimental parameters It allows surface phenomena to be probed in more than

one way in order to eliminate ambiguities andor to reinforce conclusions It

eliminates the uncertainties that arise from data taken at different locations

using the techniques separately hence it allows better correlations between

data taken using the different techniques

In this study several aspects of the chemisorption interaction of

hydrogen with the low index (100) plane of single crystal tungsten denoted

3

W(100) have been investigated The choice of the tungsten-hydrogen system is

motivated by several factors The literature raises questions on the unusual

state behavior observed when hydrogen adsorbs on W(100) questions which our

apparatus is especially well-suited to study In addition the interaction of

hydrogen with close packed planes of single crystal metals including W(100)

is of particular interest to theoreticians since hydrogen is the simplest

molecular gas as detailed in Sections 128 and 129 below recent theories

of chemisorption have been applied to the hydrogen-tungsten system Tungsten

was chosen as the substrate metal because of its relative ease of preparation

and cleaning and because the large body of literature on the hydrogen-tungsten

system indicates wide interest in this interaction

In particular this research is concentrated in three areas Tung~

sten when cleaned and subsequently cooled to below room temperature exhibits

an anomalous centered (2x2) denoted C(2x2) low energy electron diffraction

pattern The possibility that this is caused by hydrogen diffusing from the

bulk or adsorbing from the background has been investigated Secondly the

room temperature and low temperature hydrogen adsorption binding states have

been studied using electron stimulated desorption and electron spectroscopy

Finally low energy electron diffraction has been used in an investigation of

the atomic geometry of the clean tungsten surface and the geometry of one of

the hydrogen binding states

The remaining sections of this chapter summarize literature relating

to this research Chapter 2 details the construction and performance of the

experimental apparatus A presentation of the experimental data is made in

Chapter 3 A discussion of the results is made in Chapter 4 A summaryof

4

the research and conclusions from it are given in Chapter 5

12 Review of Literature

121 Introduction

Experimental investigation of the adsorption of hydrogen on tungsten

has been carried out using field emitter tips macroscopic polycrystalline

surfaces and all the low index orientation single crystal faces including

W(lOO) From the wide range of results it is clear that the substrate exerts

a strong influence on the adsorption properties The question has been raised

whether adsorption models postulated from polycrystalline substrates can be

3extended even qualitatively to single crystal systems Likewise there are

such sharp differences in the chemisorption behavior of hydrogen onto the

different single crystal tungsten surfaces that the specific physical process

governing adsorption on each face appears to be unique to that face This

review will therefore be concerned for the most part with literature dealing

with hydrogen adsorption on W(lOO) the gas-substrate system used in this

research

The following data review is presented in subsections each concenshy

trating on a specific experimental technique or class of experiments Subsecshy

tion 127 includes some gas mixing literature which was considered relevant

and Subsection 129 gives a bibliography of some related chemisorption

theoretical papers

122 Thermal Desorption

One technique widely employed in investigation of the kinetics of gas

456adsorption has been thermal or flash desorption By monitoring the system

5

pressure changes of a species while rapidly heating the crystal inferences

can be made on the existence of binding states adsorption energies and reaction

7 8 order Heating is accomplished by a focussed light beam electron bombardshy

7 9ment or ohmic losses bull

The strongest indication that hydrogen exhibits multiple state

adsorption behavior on W(lOO) comes from flash desorption data Tamm and

8Schmidt 7 and Madey and Yates found for room temperature adsorption that two

large peaks are observed in the plot of pressure of H2 versus temperature

These peaks are centered at 4500 K and approximately 550oK Only hydrogen

molecules are observed to be desorbed from the surface Using desorption rate

456theory the lower temperature (~l) state is found to obey first order

kinetics the peak temperature being independent of initial coverage The

higher temperature (~2) state obeys second order kinetics the peak temperature

moving down from approximately 5800 K to 5500 K as the initial coverage is

increased

In addition to the two main peaks several small peaks comprising a

total of less than 10 percent of the desorbed hydrogen are observed during a

flash when the crystal is cooled to nitrogen temperature before hydrogen

exposure These states are believed to result from adsorption onto edges and

other non-(lOO) crystal orientations 8

The dependence of the desorption energy Ed on coverage is weak

~l 8 ~2 8with Ed = 252 kcallmole and Ed = 323 kcalmole As would be predicted

from this result the ~2 state is largely populated before the ~l state begins

to fill The ratio of the populations in the states at full coverage n~1n~2

8 lOis 2111 Jelend and Menzel point out that the cross section for electron

6

desorption (see Subsection 125) from the ~1 state is considerably smaller

than that of the ~2 state a result not predicted from the experimental flash

desorption binding energies

11Tibbetts using a stepped temperature flash desorption technique

has shown that hydrogen is bound to the surface of po1ycrysta11ine tungsten

W(110) and W(lll) with a distribution of binding energies ranging from

162 kca1mo1e (07 eV) to 415 kca1mo1e (lS eV) Using this technique

desorption at a constant temperature from a distribution of states can be easily

distinguished from desorption from a state with a unique binding energy While

he did not extend his work to W(100) these results suggest that hydrogen does

not bind to the surface at a few energies but rather with a range of binding

energies Tibbetts also shows that the determination of the order of the

desorption process is difficult for adsorption into a range of binding energies

In experiments using a combination of hydrogen and deuterium as room

temperature adsorbates the elemental composition of the desorbed ~1 and ~2

states shows complete isotopic mixing between the states7S independent of the

adsorption sequence or ratios However for low temperature adsorption

complete state conversion is not observed for all the additional small peaks

Where mixing is observed atomic adsorption is suggested Where mixing is not

observed molecular adsorption is inferred

Some theoretical models indicate that multiple peak flash desorption

spectra may be due to binding energy changes caused by adsorbate-adsorbate

interactions12 or order-disorder transactions during f1ash3 13 rather than

actual distinct multiple binding states under the adsorption conditions

7

123 LEED and HEED

Low energy electron diffraction or LEED refers to the coherent

scattering of a monoenergetic focused electron beam from the surface of a

crystalline solid Electrons in the energy range 0 lt E lt 300 eV interact

strongly with the ion cores and valence electrons of the solid The inelastic

mean free path is on the order of 10 X As a consequence electrons which are

observed elastically have been backscattered within the first several surface

layers in the solid making LEED a surface sensitive probe 14 15 The periodic

translational symmetry of the surface leads to the momentum conservation laws

I

= (11)~I ~II + g

= (2mEh2)t sin reg (12)~II

where ~II is the momentum parallel to the surface of the emerging electron and

~ is the parallel momentum of the incident electron The vector g is a

reciprocal lattice vector of the solid surface and reg is the polar angle of the

incident beam As a result of the conservation laws the scattered electrons

emerge in a series of beams called a spot pattern which is a manifestation

of the surface symmetry The beams are labelled by their corresponding g

vectors Information on the actual configuration of the surface ion cores is

extracted from the plot of elastically scattered electron intensity versus

16 17incident energy called elastic intensity profiles or I-V curves

Changes in the LEED spot pattern resulting from hydrogen adsorption

18onto W(lOO) at room temperature have been reported by Estrup and Anderson and

19by Yonehara and Schmidt As hydrogen is adsorbed at room temperature the

8

(lxl) pattern characteristic of clean tungsten changes into a weak C(2x2)

pattern at a hydrogen coverage e of less than 005 monolayer (where e = 1

15 2monolayer at a coverage of 2xlO hydrogen atomsem which is 2 hydrogen atoms

for each surface tungsten atom) The (1212) spots become sharp and intense

with increased hydrogen adsorption reaching a maxima in intensity at approxishy

19 18mately 015 monolayer or e = 025 monolayer (Coverage in both cases was

determined by flash desorption) The (1212) spots split into four new spots

which move apart continuously and upon further hydrogen adsorption eventually

19form 13-order streaks at a coverage of ~ 035 monolayer These in turn

change at higher coverage into poorly developed 14-order features with the

(012) spots missing Finally near e = 10 all extra diffraction features

are gone and the LEED pattern returns to a (lxl) symmetry This series of LEED

pattern changes is reversible with coverage

For hydrogen adsorption at 1950 K the C(2x2) pattern develops

maximizes in intensity at e ~ 18 and splits into four beams similar to room

temperature adsorption However the higher coverage LEED structures third

19 0order streaks and fourth order spots are not observed At 78 K adsorption

of hydrogen produces half-order C(2x2) spots that maximize in intensity at

e ~ 027 Additional hydrogen adsorption beyond this causes the half-order

19spot intensity to decrease but with no splitting of the spots

If clean W(lOO) is allowed to cool below room temperature a sharp

19and intense C(2x2) pattern is observed even though total coverage due to the

background of H2 and CO is less than e = 001 as determined by flash desorption

measurements The intensity of these half-order beams can be altered by

additional hydrogen adsorption or by various hydrogen and oxygen heat treatmenm

9

l9Yonehara and Schmidt have postulated that the extra spots may result from

hydrogen endothermally soluble in tungsten diffusing to the surface Possible

alternative explanations include the reconstruction of the clean tungsten

surface or impurity over layer adsorption from the background It should be

20noted that Plummer and Bell saw no effects upon cooling their field emitter

tip to 78 0 K that they could attribute to hydrogen diffusion to the surface

Elastic intensity profiles for several beams have been measured by

18 21Estrup and Anderson and by Papageorgopoulos and Chen at room temperature

Estrup and Anderson reported profiles for (00) and (1212) tEED lJeams out

to 300 eV The intensity variation for the (OO) beam was found to be indepenshy

dent of hydrogen coverage a manifestation of the relatively weak scattering

cross section of a hydrogen ion core relative to that of tungsten The

additional beams observed on adsorption would be expected to be more sensitive

13to the adsorbed overlayer Estrup found the intensity of the (1212) beams

for H W(100) and D W(100) to decrease as a function of temperature between2 2

300o

K and 6000 K at a constant incident energy but neither obeyed the Debyeshy

22Waller relation Estrup interpreted this result to imply the adsorbed

species are directly involved in the diffraction process

19Yonehara examined the tEED spot intensities as a function of

temperature down to liquid nitrogen temperatures His results are examined

in the data discussion in Chapter 4

The hydrogen-WlOO) system has also been studied using high energy

electron diffraction or HEED23 This technique utilizes the strong forward

scattering of high energy electrons in probing the surface by having the

incident beam strike the solid at grazing incidence The resulting diffraction

10

patterns are sensitive to the translational symmetry of the surface As in

LEED a diffraction pattern with C(2x2) symmetry is observed during initial

hydrogen adsorption With further hydrogen exposure the beams broaden some

splitting is observed and eventually the half-order intensity decreases

completely The width of the observed half-order beams has been interpreted to

indicate hydrogen initially adsorbs into a large number of small square or

rectangular domains each consisting of 20 to 25 atoms with sides parallel to

the [100J directions

124 Electron Scattering Field Emission and Photoemission

The electronic structure of the surface layers of solids can be

characterized by electron emission spectroscopies Such techniques require a

surface sensitive perturbation on the gas-solid system leading to an electron

yield A spectrometer is used to energy analyze the emitted or scattered

24 25electrons In electron scatter~ng a monoenergetic beam of electrons is

directed onto a crystal The inelastically scattered signal containing

quantitative information about electronic excitations at the surface is

measured as a function of energy loss Field emission is a technique in which

application of a high positive electric field to the surface lowers the potenshy

26tia1 barrier for electrons to tunnel out of the system The total energy

distribution of electrons emitted as a result of the applied field contains

bull 27 28spectroscopic data on the adsorbed spec~es Structure induced by the

adsorbate in the elastic tunneling spectrum is thought to be a measure of the

d f i h f h d b 2930e1ectron~c ens~ty 0 states n t e v~c~n~ty 0 tea sor ate The

inelastic tunneling spectrum reveals discrete excitation energies in the

11

gas-surface system In photoemission or ftotoelectron spectroscopy an incident

monochromatic light beam incident on the solid induces emission of electrons by

interband transitions with or without an inelastic scattering after the initial

31 exc i tat~on

The vibrational spectrum of hydrogen adsorbed on the tungsten (100)

32surface has been observed by Propst and Piper using inelastic low energy

20electron scattering and by Plummer and Bell using field emission Propst

and Piper found two peaks in the energy loss distribution E - E pr~mary scattered

at 135 meV and 69 meV for all coverages used up to saturation These results

are interpreted as indicating hydrogen atoms are multiply bonded to tungsten

33atoms No losses near 550 meV the vibrational energy of molecular hydrogen

were observed Plummer and Bell measured the total energy distribution of

field emitted electrons from tungsten (100) as a function of coverage of

hydrogen and deuterium at several temperatures Discrete losses at 100 meV

and 50 meV for deuterium which correspond to the 135 meV and 69 meV vibrational

losses observed by Propst and Piper were evident for deuterium in the a state2

No discrete energy losses at 550 meV or 390 meV the molecular vibrational

levels of H2 and D2 respectively were measured Hence there was no observable

molecular adsorption at room temperature

Additional structure in the field emission total energy distribution

is observed as a function of hydrogen adsorption The clean surface spectrum

has two peaks 04 eV and 15 eV down from the Fermi level which have been

d ifi d h h d f d h h f 34 A h d ~ ent e as ~g ens~ty 0 states assoc~ate w~t t e sur ace s y rogen

(or deuterium) coverage increases between 0 and 5xl014 atomscm2

ie while

the ~2 state is filling the surface states disappear and the field emitted

12

energy distribution changes from that of clean tungsten to a spectrum indicative

of resonance tunneling through a level positioned 09 eV below the Fermi energy

There is a shoulder at -11 eV and an energy dependent background With

further adsorption up to saturation this energy distribution characteristic

of the ~2 state shifts to a broad level centered at 26 eV below the Fermi

level This is interpreted to indicate that the ~1 state does not adsorb on

top of the ~2 state

35The photoemission results of Wac1awski and P1ummer and Feuerbacher

36and Fitton are in agreement with those obtained using the field emission

technique As hydrogen adsorbs on the surface emission from the 04 eV

surface state decreases The level 09 eV down from the Fermi energy as well

as a level 5 eV down remain unchanged by adsorption At full coverage a strong

peak at -25 eV is observed

125 Electron Stimulated Desorption

Desorption of energetic ions and neutrals from a gas over layer can

be induced by electron bombardment This process termed electron stimulated

desorption or ESD37383940 is described below in the discussion of the data

in Chapter 4 In brief intensity and energy measurements of the ion yield can

be interpreted in terms of the parameters characterizing the binding energy

curve of the adsorbed state and adsorption states can be classified by cross

section for desorption_ ion energy distributions primary energy threshold for

desorption etc

ESD has been employed to study adsorption kinetics for the hydrogenshy

41 ~ tungsten (100) system Madey obtained a coverage dependence yield of H ions

13

observing an ion signal during the initial stages of hydrogen adsorption that

reached a maximum at a coverage of e = 017 and then decayed to zero yield

23 2after e = 075 the maximum cross section obtained was 18x10- cm at a

primary electron energy of 100 eV Tamm and Schmidt7 reported a maximum cross

section for H+ desorption of 10- 21 cm 2bull Benninghoven Loebach and Treitz42

10and Je1end and Menze1 obtained qualitatively similar (to Madey) coverage

dependencies on tungsten ribbons which were believed to be predominantly of

(100) crystallites the main difference between these results and Madeys is

the observation of a residual signal at high hydrogen coverage

The cross section results indicate that the a state even though it2

is more tightly bound than the a state provides more hydrogen ions upon1

electron impact The cross section for ion desorption from the a1 state is

d41 10-25est1mate to be 1ess t han cm2 bull

+No energy distributions of the desorbed H ions have been reported

The dependence of the desorption cross section on the primary electron energy

10has not been investigated Je1end and Menzel found their results independent

of temperature for temperatures above room temperature No ESD results for

otemperatures less than 300 K have been reported

126 Work Function

The work function is the minimum amount of energy required to remove

43 an electron from a meta1 An adsorbate may alter the potential barrier for

electrons at the surface and hence change the work function Such is the case

for hydrogen on W(100) Hydrogen adsorbs as an electronegative ion on W(100)

thereby increasing the work function

14

44Using a retarding potential current-voltage techn~que Madey and

8Yates reported a linear relation between work function change and hydrogen

coverage over the entire coverage range for adsorption occurring at a

temperature 0 f 2= 300 0 K Coverages were determined by flash desorption

Total work function change at one monolayer was found to be 090 eV which

corresponds to a constant dipole moment per adsorbed atom of 015 Debye

15 2(assuming a total coverage of 15xlO atomsem)

Estrup and Anderson18 using the retarding potential technique and

Plummer and Bel120 using a field emission Fowler-Nordheim plot45 found a

work function change versus coverage relationship which deviated only slightly

8from being linear in near agreement with the results of Madey and Yates

46 47Becker using field emission and Armstrong using a retarding potential

technique found a work function change for the H W(100) system of approximately2

48090 eV when e = 10 Hopkins and Pender obtained an anomalously low value

49of 054 eV while Collins and Blott obtained a high value of 109 eV for the

same coverage

127 Chemical Probes

Hydrogen adsorption has also been studied using a chemical probe

ie by coadsorbing some additional different gas with the hydrogen and

observing the effect this additional species had on results obtained using

techniques discussed in previous sections

50Yates and Madey found using flash desorption that CO coadsorption

onto a hydrogen monolayer displaces both the Sl and S2 binding states with

near unity probability in doing so initially replacing these states with a

15

less tightly bound hydrogen state After high CO exposure no hydrogen remains

on the surface if the coadsorption is done at room temperature At lower

temperatures only about 10 percent of the hydrogen is desorbed Jelend and

Menzel lO 5l found that the ionic and total desorption cross sections for

hydrogen the sticking probability and bond strength of hydrogen are all

affected by preadsorption of CO or oxygen Yates and Madey3 found that N2

slowly replaces preadsorbed hydrogen at temperatures greater than 300oC Below

2730

K the displacement of hydrogen by N2 does not occur

These results all show that small impurity concentrations of CO N

and 0 if intentionally introduced do affect the resulting hydrogen chemishy

sorption properties These experiments underscore the necessity of surface

chemical characterization

128 Chemisorption Models

7 18 19 52 Several models have been proposed to account for the

experimental results from hydrogen chemisorption on W(lOO) Each of these

models embodies some combination of the following three phenomenological

assumptions 1) The tungsten surface is ideal planar and free of contamishy

nants 2) There are available two types of adsorbates namely hydrogen atoms

53and molecules 3) There are three high symmetry logical bonding sites on

a BCC (100) face usually designated by the letters A B and C These

positions are A a bridge site with 2 nearest neighbors and the hydrogen

equidistant from two adjacent surface tungsten atoms B a four fold symmetric

site with 5 nearest neighbors and the hydrogen located on a surface normal that

passes through the center of a surface primitive unit cell and C the hydrogen

16

located directly above a surfac~ atom with one nearest neighbor There are

15 152x10 A sitescm2 and 1x10 B ore sites~m2 on middotW(100)

The proposed models attempt to explain or at least be consistent

with empirical results presented in Sections 121 through 127 A summary

of salient features of the reported research includes a) a linear work function

15change with coverage b) a saturation coverage of between 15 and 20x10 H

2atomscm c) a coverage dependent adsorption layer symmetry including a we11shy

ordered surface at low coverages d) the appearance at low temperatures of a

surface translational symmetry not equivalent to room temperature clean

tungsten e) adsorption into states which produces a two state thermal desorpshy

tion spectrum with relative coverages in the two states of nQ n = 21 f) a ~1 ~2

coverage dependence for the observed electronic structure including the

absence at high coverage of structure observed at low coverages and g) no

observed vibrational levels which correspond to free H2 molecule

Following is a short review of several proposed surface models

18Estrup and Anderson account for their LEED spot pattern results

using a model wherein the hydrogen molecules dissociate upon adsorption and

migrate initially to form islands of hydrogen atoms located at type A sites in

a e(2x2) structure Dissociation is assumed since the e(2x2) pattern maximizes

15 2at a coverage of 25x10 mo1ecu1escm whereas without dissociation it would

maX1m1ze at 5 1015 mo ecu1es cm2bull e site adsorptionx 1 Although type B or type

could also produce the initial e(2x2) LEED pattern these sites are rejected

15 2because of the following 1) There are only 1x10 B or e sitescm This is

not enough sites to accommodate a full monolayer coverage if all binding is

atomic and if it is into equivalent sites as suggested by the linear work

17

function change with coverage 2) While a complete 14-order LEED pattern is

possible with type B or type C sites it is not possible to produce a 14-order

54pattern with (012) spots missing using B or C sites As the adsorption

14 2proceeds past a surface density of 5xlO hydrogen atomscm the adatoms conshy

tinuously rearrange into long period superlattices with antiphase domains

that produce successively the 12-order spot splitting the 13-order streak

(only partly ordered domains) the 14-order pattern and finally the (lxl)

full coverage structure

This model has been criticized for not accounting for the two

observed binding states and for predicting a maximum in intensity of the half

order spots at e = 025 these beams maximize at less than that coverage In

addition the authors themselves questioned the idea of well-ordered super-

lattices

In order to explain the relative intensity of the two states at

saturation seen in flash desorption n~~ = 21 Tamm and Schmidt7

propose a 1 2

model wherein the initial hydrogen adsorption is atomic and at B sites in a

C(2x2) pattern At full coverage the remaining B sites are filled with

hydrogen molecules B sites are chosen over C sites using molecular orbital

55arguments A sites are rejected as there are too many to satisfy this model

This model satisfies the second order kinetics observed for the ~2 state and

the first order kinetics observed for the ~l state In addition it explains

14 2 7 their saturation coverage of gt 3xl0 moleculescm One weakness of this

model pointed out by Tamm and Schmidt is that it is not obvious the model can

account for the observed LEED and work function results Also no molecular

vibrations have been observed

18

19yonehara and Schmidt propose a three bonding state model that is

more consistent with the low temperature LEED results In their model an

additional state labeled ~3 is always present at the surface in A or B

2sites with a density of 5xl014 atomscm bull This state orders on the surface at

low temperatures to give the C(2x2) pattern observed At room and higher

temperatures it goes into solution (or perhaps disorders) becoming invisible

to LEED In the type A site model adsorption from the background fills in the

remaining A sites sequentially with the ~2 and ~l states This model is similar

18to that proposed by Estrup and Anderson except that the ~3 state although

invisible to LEED at room and higher temperatures occupies 14 of the type A

sites This model eliminates the stoichiometric objection to Estrup and

Andersons model The model depends on the possibility of type A sites being

occupied by states with considerably different properties If the ~3 state

occupies type B sites ~2 state adsorption fills in the remaining type B sites

and then ~l hydrogen adsorbs into random or possibly type C sites This

second model does not explain satisfactorily work function or LEED spot pattern

experimental data

56Propst found using a point defect type calculation that only a

B-site model of adsorption would produce the observed vibrational energies of

5769 meV and 135 meV Park using LEED intensity arguments found that only

adsorption into sites with 4-fold symmetry on the (100) surface (type B or type

C sites) would be consistent with broadening and splitting of the half-order

beams while the integral beams remain sharp

8Madey and Yates offer a critique of several of the proposed models

bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

iv

TABLE OF CONTENTS

Chapter Page

1 MOTIVATION AND REVIEW OF LITERATURE bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 1

11 Introduction and Motivationbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 1

12 Review of Literature bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 4o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

121 Introductionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 4 122 Thermal Desorptionbull o 4 123 LEED and HEEDbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 7 124 Electron Scattering Field Emission and

Photoemissionbullbullbull 100 bullbullbullbullbull 0 bullbullbullbull

125 Electron Stimulated Desorptionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 12 126 Work Functionbullbullbullbullbullbullbullbullbull 13o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

1 27 Chemical Probes bullbullbullbullbullbullbull bullbullbullbull bullbullbullbull 14I) O

128 Chemisorption Models bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 15 1 29 Theoretical Treatment bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 19 1 210 Literature Surnmaryo 20

13 Review of Experimental Techniques bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 21

2 EXPERIMENTAL APPARATUS bullbullbullbullbullbullbullbullbullbullbullbull ~ bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 22

21 CGeneral Description bull 22

22 Vacuum System and Magnetic Shieldingbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 22

23 Electron Gun and Gun Electronics bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 25

24 Energy Spectrometer 29

25 Spectrometer Electronics bullbull 45o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

26 LEED Display System 490 bullbullbull 0 bullbullbullbullbullbullbull 0 bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull I)

27 Crystal Manipulator Assembly bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 51

28 Crystal Preparation and Cleaning Procedures bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 57

29 Gas Handling System o bullbullbullbullbullbullbullbullbullbullbullbullbull o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 62

210 Mass Spectrometer bullbullbullbull o 64

3 EXPERIMENTAL PROCEDURE AND RESULTS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 65

31 Introductionbullbullbullbullbullbullbullbullbullbullbullbull 65

32 Experimental Procedure o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull o bullbullbullbullbullbullbull 65 321 Work Function and Coverage Measurements bullbullbullbullbullbullbullbullbullbullbullbullbull 65 322 Primary Electron Beam Energy and Current

Measurements 6811 bullbullbullbullbullbullbullbullbullbullbullbullbullbullbull bullbullbullbull

323 Target Temperature Measurement bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 69 324 Electron Scattering and Auger Measurements bullbullbullbullbullbullbullbullbullbull 69 325 Ion Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 72o bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull

v

33 Ion Species Identificationbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 74

34 Work Function Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 77

35 u+ Ion Desorption Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 83 351 ut Ion Energy Distributionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 83 352 Desorption Curve Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 93 353 Temperature Dependence of the Fast State a+ Ion

Signal 4) 96

354 Additional Measurements on the Fast State bullbullbullbullbullbullbullbullbullbull 98

36 Low Energy Electron Diffraction Data bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull lll

37 Electron Scattering Data bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 119

4 DISCUSSION OF RESULTS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 126

41 Auger Electron Spectroscopy Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 126

42 Work Function Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 127

43 Electron Stimulated Desorption Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 129 431 Room Temperature Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 130 432 Low Temperature Fast State Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 135 433 Low Temperature Dosed Surface Results bullbullbullbullbullbullbullbullbullbullbullbullbull142

44 Low Energy Electron Diffraction Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 145

45 Electron Scattering Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 161

5 SUMMARY AND CONCLUSIONS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull165

REFERENCES 171

APPENDIX A 0 176 poundI

APPENDIX B 9 c 179

APPENDIX C 0 182 0 bull

VITA 190lilt bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull bullbull

1

CHAPTER 1

MOTIVATION AND REVIEW OF LITERATURE

11 Introduction and Motivation

The experimental investigation of chemisorption of gases onto metallic

surfaces has uncovered in the past decade a baffling complexity of adsorption

geometries and binding state kinetics and energetics A growing interest in

surface phenomena generated by the questions raised in these researches and

by the long anticipated use of the answers has led to the rapid development

of new surface probing techniques and to an increase in the availability of new

technologies compatible with surface research Aided by the new technologies

and techniques chemisorption research has been undertaken on many different

systems (system =solid substrate + gas adsorbate) under a wide range of

environmental conditions Surface theoreticians have proposed microscopic and

phenomenological models with which data have been analyzed and compared A

large experimental and theoretical literature in the field has resulted

The study of adsorption on metal surfaces has among its primary

objectives a detailed description of the physical geometry of the atoms in the

surface layers and a quantitative understanding of electronic and chemical

surface phenomena While considerable progress has been made in documenting

gas-surface phenomena it has been only in the past three years that any

credible claims to understanding specific surface overlayer geometries have

1been made At the same time many basic questions involving system kinetics

and energetics have not been adequately resolved 2

It has become apparent from the failure of the research effort to

satisfy the above objectives and from seemingly conflicting data that adequate

2

characterization is basic to all surface experimentation The need for such

characterization has been strongly emphasized by disagreement between quasishy

similar research performed at different locations Small impurity concentrashy

tions small changes in crystal orientation minor temperature differences etc

often have considerable effects on experimental measurements

A corollary to the necessity of being able to characterize the

surface is the importance of being able to control the surface characterization

In order to compare data it is necessary to reproduce all experimental paramshy

eters accurately The theory to which experimental data are compared is

constrained by complexity to be based on ideal clean substrates with a

definable gas-surface interaction It is desirable to be able to choose the

adsorbate-substrate system and experimental parameters that can also be used

with theoretical models Of course compromises are inevitable

One approach in surface research has been to incorporate several

experimental techniques into one apparatus At the present time this approach

is increasingly more common A combination of surface sensitive probes offers

many advantages It provides for more checks of the data against results

published in the literature It allows for better characterization of the

experimental parameters It allows surface phenomena to be probed in more than

one way in order to eliminate ambiguities andor to reinforce conclusions It

eliminates the uncertainties that arise from data taken at different locations

using the techniques separately hence it allows better correlations between

data taken using the different techniques

In this study several aspects of the chemisorption interaction of

hydrogen with the low index (100) plane of single crystal tungsten denoted

3

W(100) have been investigated The choice of the tungsten-hydrogen system is

motivated by several factors The literature raises questions on the unusual

state behavior observed when hydrogen adsorbs on W(100) questions which our

apparatus is especially well-suited to study In addition the interaction of

hydrogen with close packed planes of single crystal metals including W(100)

is of particular interest to theoreticians since hydrogen is the simplest

molecular gas as detailed in Sections 128 and 129 below recent theories

of chemisorption have been applied to the hydrogen-tungsten system Tungsten

was chosen as the substrate metal because of its relative ease of preparation

and cleaning and because the large body of literature on the hydrogen-tungsten

system indicates wide interest in this interaction

In particular this research is concentrated in three areas Tung~

sten when cleaned and subsequently cooled to below room temperature exhibits

an anomalous centered (2x2) denoted C(2x2) low energy electron diffraction

pattern The possibility that this is caused by hydrogen diffusing from the

bulk or adsorbing from the background has been investigated Secondly the

room temperature and low temperature hydrogen adsorption binding states have

been studied using electron stimulated desorption and electron spectroscopy

Finally low energy electron diffraction has been used in an investigation of

the atomic geometry of the clean tungsten surface and the geometry of one of

the hydrogen binding states

The remaining sections of this chapter summarize literature relating

to this research Chapter 2 details the construction and performance of the

experimental apparatus A presentation of the experimental data is made in

Chapter 3 A discussion of the results is made in Chapter 4 A summaryof

4

the research and conclusions from it are given in Chapter 5

12 Review of Literature

121 Introduction

Experimental investigation of the adsorption of hydrogen on tungsten

has been carried out using field emitter tips macroscopic polycrystalline

surfaces and all the low index orientation single crystal faces including

W(lOO) From the wide range of results it is clear that the substrate exerts

a strong influence on the adsorption properties The question has been raised

whether adsorption models postulated from polycrystalline substrates can be

3extended even qualitatively to single crystal systems Likewise there are

such sharp differences in the chemisorption behavior of hydrogen onto the

different single crystal tungsten surfaces that the specific physical process

governing adsorption on each face appears to be unique to that face This

review will therefore be concerned for the most part with literature dealing

with hydrogen adsorption on W(lOO) the gas-substrate system used in this

research

The following data review is presented in subsections each concenshy

trating on a specific experimental technique or class of experiments Subsecshy

tion 127 includes some gas mixing literature which was considered relevant

and Subsection 129 gives a bibliography of some related chemisorption

theoretical papers

122 Thermal Desorption

One technique widely employed in investigation of the kinetics of gas

456adsorption has been thermal or flash desorption By monitoring the system

5

pressure changes of a species while rapidly heating the crystal inferences

can be made on the existence of binding states adsorption energies and reaction

7 8 order Heating is accomplished by a focussed light beam electron bombardshy

7 9ment or ohmic losses bull

The strongest indication that hydrogen exhibits multiple state

adsorption behavior on W(lOO) comes from flash desorption data Tamm and

8Schmidt 7 and Madey and Yates found for room temperature adsorption that two

large peaks are observed in the plot of pressure of H2 versus temperature

These peaks are centered at 4500 K and approximately 550oK Only hydrogen

molecules are observed to be desorbed from the surface Using desorption rate

456theory the lower temperature (~l) state is found to obey first order

kinetics the peak temperature being independent of initial coverage The

higher temperature (~2) state obeys second order kinetics the peak temperature

moving down from approximately 5800 K to 5500 K as the initial coverage is

increased

In addition to the two main peaks several small peaks comprising a

total of less than 10 percent of the desorbed hydrogen are observed during a

flash when the crystal is cooled to nitrogen temperature before hydrogen

exposure These states are believed to result from adsorption onto edges and

other non-(lOO) crystal orientations 8

The dependence of the desorption energy Ed on coverage is weak

~l 8 ~2 8with Ed = 252 kcallmole and Ed = 323 kcalmole As would be predicted

from this result the ~2 state is largely populated before the ~l state begins

to fill The ratio of the populations in the states at full coverage n~1n~2

8 lOis 2111 Jelend and Menzel point out that the cross section for electron

6

desorption (see Subsection 125) from the ~1 state is considerably smaller

than that of the ~2 state a result not predicted from the experimental flash

desorption binding energies

11Tibbetts using a stepped temperature flash desorption technique

has shown that hydrogen is bound to the surface of po1ycrysta11ine tungsten

W(110) and W(lll) with a distribution of binding energies ranging from

162 kca1mo1e (07 eV) to 415 kca1mo1e (lS eV) Using this technique

desorption at a constant temperature from a distribution of states can be easily

distinguished from desorption from a state with a unique binding energy While

he did not extend his work to W(100) these results suggest that hydrogen does

not bind to the surface at a few energies but rather with a range of binding

energies Tibbetts also shows that the determination of the order of the

desorption process is difficult for adsorption into a range of binding energies

In experiments using a combination of hydrogen and deuterium as room

temperature adsorbates the elemental composition of the desorbed ~1 and ~2

states shows complete isotopic mixing between the states7S independent of the

adsorption sequence or ratios However for low temperature adsorption

complete state conversion is not observed for all the additional small peaks

Where mixing is observed atomic adsorption is suggested Where mixing is not

observed molecular adsorption is inferred

Some theoretical models indicate that multiple peak flash desorption

spectra may be due to binding energy changes caused by adsorbate-adsorbate

interactions12 or order-disorder transactions during f1ash3 13 rather than

actual distinct multiple binding states under the adsorption conditions

7

123 LEED and HEED

Low energy electron diffraction or LEED refers to the coherent

scattering of a monoenergetic focused electron beam from the surface of a

crystalline solid Electrons in the energy range 0 lt E lt 300 eV interact

strongly with the ion cores and valence electrons of the solid The inelastic

mean free path is on the order of 10 X As a consequence electrons which are

observed elastically have been backscattered within the first several surface

layers in the solid making LEED a surface sensitive probe 14 15 The periodic

translational symmetry of the surface leads to the momentum conservation laws

I

= (11)~I ~II + g

= (2mEh2)t sin reg (12)~II

where ~II is the momentum parallel to the surface of the emerging electron and

~ is the parallel momentum of the incident electron The vector g is a

reciprocal lattice vector of the solid surface and reg is the polar angle of the

incident beam As a result of the conservation laws the scattered electrons

emerge in a series of beams called a spot pattern which is a manifestation

of the surface symmetry The beams are labelled by their corresponding g

vectors Information on the actual configuration of the surface ion cores is

extracted from the plot of elastically scattered electron intensity versus

16 17incident energy called elastic intensity profiles or I-V curves

Changes in the LEED spot pattern resulting from hydrogen adsorption

18onto W(lOO) at room temperature have been reported by Estrup and Anderson and

19by Yonehara and Schmidt As hydrogen is adsorbed at room temperature the

8

(lxl) pattern characteristic of clean tungsten changes into a weak C(2x2)

pattern at a hydrogen coverage e of less than 005 monolayer (where e = 1

15 2monolayer at a coverage of 2xlO hydrogen atomsem which is 2 hydrogen atoms

for each surface tungsten atom) The (1212) spots become sharp and intense

with increased hydrogen adsorption reaching a maxima in intensity at approxishy

19 18mately 015 monolayer or e = 025 monolayer (Coverage in both cases was

determined by flash desorption) The (1212) spots split into four new spots

which move apart continuously and upon further hydrogen adsorption eventually

19form 13-order streaks at a coverage of ~ 035 monolayer These in turn

change at higher coverage into poorly developed 14-order features with the

(012) spots missing Finally near e = 10 all extra diffraction features

are gone and the LEED pattern returns to a (lxl) symmetry This series of LEED

pattern changes is reversible with coverage

For hydrogen adsorption at 1950 K the C(2x2) pattern develops

maximizes in intensity at e ~ 18 and splits into four beams similar to room

temperature adsorption However the higher coverage LEED structures third

19 0order streaks and fourth order spots are not observed At 78 K adsorption

of hydrogen produces half-order C(2x2) spots that maximize in intensity at

e ~ 027 Additional hydrogen adsorption beyond this causes the half-order

19spot intensity to decrease but with no splitting of the spots

If clean W(lOO) is allowed to cool below room temperature a sharp

19and intense C(2x2) pattern is observed even though total coverage due to the

background of H2 and CO is less than e = 001 as determined by flash desorption

measurements The intensity of these half-order beams can be altered by

additional hydrogen adsorption or by various hydrogen and oxygen heat treatmenm

9

l9Yonehara and Schmidt have postulated that the extra spots may result from

hydrogen endothermally soluble in tungsten diffusing to the surface Possible

alternative explanations include the reconstruction of the clean tungsten

surface or impurity over layer adsorption from the background It should be

20noted that Plummer and Bell saw no effects upon cooling their field emitter

tip to 78 0 K that they could attribute to hydrogen diffusion to the surface

Elastic intensity profiles for several beams have been measured by

18 21Estrup and Anderson and by Papageorgopoulos and Chen at room temperature

Estrup and Anderson reported profiles for (00) and (1212) tEED lJeams out

to 300 eV The intensity variation for the (OO) beam was found to be indepenshy

dent of hydrogen coverage a manifestation of the relatively weak scattering

cross section of a hydrogen ion core relative to that of tungsten The

additional beams observed on adsorption would be expected to be more sensitive

13to the adsorbed overlayer Estrup found the intensity of the (1212) beams

for H W(100) and D W(100) to decrease as a function of temperature between2 2

300o

K and 6000 K at a constant incident energy but neither obeyed the Debyeshy

22Waller relation Estrup interpreted this result to imply the adsorbed

species are directly involved in the diffraction process

19Yonehara examined the tEED spot intensities as a function of

temperature down to liquid nitrogen temperatures His results are examined

in the data discussion in Chapter 4

The hydrogen-WlOO) system has also been studied using high energy

electron diffraction or HEED23 This technique utilizes the strong forward

scattering of high energy electrons in probing the surface by having the

incident beam strike the solid at grazing incidence The resulting diffraction

10

patterns are sensitive to the translational symmetry of the surface As in

LEED a diffraction pattern with C(2x2) symmetry is observed during initial

hydrogen adsorption With further hydrogen exposure the beams broaden some

splitting is observed and eventually the half-order intensity decreases

completely The width of the observed half-order beams has been interpreted to

indicate hydrogen initially adsorbs into a large number of small square or

rectangular domains each consisting of 20 to 25 atoms with sides parallel to

the [100J directions

124 Electron Scattering Field Emission and Photoemission

The electronic structure of the surface layers of solids can be

characterized by electron emission spectroscopies Such techniques require a

surface sensitive perturbation on the gas-solid system leading to an electron

yield A spectrometer is used to energy analyze the emitted or scattered

24 25electrons In electron scatter~ng a monoenergetic beam of electrons is

directed onto a crystal The inelastically scattered signal containing

quantitative information about electronic excitations at the surface is

measured as a function of energy loss Field emission is a technique in which

application of a high positive electric field to the surface lowers the potenshy

26tia1 barrier for electrons to tunnel out of the system The total energy

distribution of electrons emitted as a result of the applied field contains

bull 27 28spectroscopic data on the adsorbed spec~es Structure induced by the

adsorbate in the elastic tunneling spectrum is thought to be a measure of the

d f i h f h d b 2930e1ectron~c ens~ty 0 states n t e v~c~n~ty 0 tea sor ate The

inelastic tunneling spectrum reveals discrete excitation energies in the

11

gas-surface system In photoemission or ftotoelectron spectroscopy an incident

monochromatic light beam incident on the solid induces emission of electrons by

interband transitions with or without an inelastic scattering after the initial

31 exc i tat~on

The vibrational spectrum of hydrogen adsorbed on the tungsten (100)

32surface has been observed by Propst and Piper using inelastic low energy

20electron scattering and by Plummer and Bell using field emission Propst

and Piper found two peaks in the energy loss distribution E - E pr~mary scattered

at 135 meV and 69 meV for all coverages used up to saturation These results

are interpreted as indicating hydrogen atoms are multiply bonded to tungsten

33atoms No losses near 550 meV the vibrational energy of molecular hydrogen

were observed Plummer and Bell measured the total energy distribution of

field emitted electrons from tungsten (100) as a function of coverage of

hydrogen and deuterium at several temperatures Discrete losses at 100 meV

and 50 meV for deuterium which correspond to the 135 meV and 69 meV vibrational

losses observed by Propst and Piper were evident for deuterium in the a state2

No discrete energy losses at 550 meV or 390 meV the molecular vibrational

levels of H2 and D2 respectively were measured Hence there was no observable

molecular adsorption at room temperature

Additional structure in the field emission total energy distribution

is observed as a function of hydrogen adsorption The clean surface spectrum

has two peaks 04 eV and 15 eV down from the Fermi level which have been

d ifi d h h d f d h h f 34 A h d ~ ent e as ~g ens~ty 0 states assoc~ate w~t t e sur ace s y rogen

(or deuterium) coverage increases between 0 and 5xl014 atomscm2

ie while

the ~2 state is filling the surface states disappear and the field emitted

12

energy distribution changes from that of clean tungsten to a spectrum indicative

of resonance tunneling through a level positioned 09 eV below the Fermi energy

There is a shoulder at -11 eV and an energy dependent background With

further adsorption up to saturation this energy distribution characteristic

of the ~2 state shifts to a broad level centered at 26 eV below the Fermi

level This is interpreted to indicate that the ~1 state does not adsorb on

top of the ~2 state

35The photoemission results of Wac1awski and P1ummer and Feuerbacher

36and Fitton are in agreement with those obtained using the field emission

technique As hydrogen adsorbs on the surface emission from the 04 eV

surface state decreases The level 09 eV down from the Fermi energy as well

as a level 5 eV down remain unchanged by adsorption At full coverage a strong

peak at -25 eV is observed

125 Electron Stimulated Desorption

Desorption of energetic ions and neutrals from a gas over layer can

be induced by electron bombardment This process termed electron stimulated

desorption or ESD37383940 is described below in the discussion of the data

in Chapter 4 In brief intensity and energy measurements of the ion yield can

be interpreted in terms of the parameters characterizing the binding energy

curve of the adsorbed state and adsorption states can be classified by cross

section for desorption_ ion energy distributions primary energy threshold for

desorption etc

ESD has been employed to study adsorption kinetics for the hydrogenshy

41 ~ tungsten (100) system Madey obtained a coverage dependence yield of H ions

13

observing an ion signal during the initial stages of hydrogen adsorption that

reached a maximum at a coverage of e = 017 and then decayed to zero yield

23 2after e = 075 the maximum cross section obtained was 18x10- cm at a

primary electron energy of 100 eV Tamm and Schmidt7 reported a maximum cross

section for H+ desorption of 10- 21 cm 2bull Benninghoven Loebach and Treitz42

10and Je1end and Menze1 obtained qualitatively similar (to Madey) coverage

dependencies on tungsten ribbons which were believed to be predominantly of

(100) crystallites the main difference between these results and Madeys is

the observation of a residual signal at high hydrogen coverage

The cross section results indicate that the a state even though it2

is more tightly bound than the a state provides more hydrogen ions upon1

electron impact The cross section for ion desorption from the a1 state is

d41 10-25est1mate to be 1ess t han cm2 bull

+No energy distributions of the desorbed H ions have been reported

The dependence of the desorption cross section on the primary electron energy

10has not been investigated Je1end and Menzel found their results independent

of temperature for temperatures above room temperature No ESD results for

otemperatures less than 300 K have been reported

126 Work Function

The work function is the minimum amount of energy required to remove

43 an electron from a meta1 An adsorbate may alter the potential barrier for

electrons at the surface and hence change the work function Such is the case

for hydrogen on W(100) Hydrogen adsorbs as an electronegative ion on W(100)

thereby increasing the work function

14

44Using a retarding potential current-voltage techn~que Madey and

8Yates reported a linear relation between work function change and hydrogen

coverage over the entire coverage range for adsorption occurring at a

temperature 0 f 2= 300 0 K Coverages were determined by flash desorption

Total work function change at one monolayer was found to be 090 eV which

corresponds to a constant dipole moment per adsorbed atom of 015 Debye

15 2(assuming a total coverage of 15xlO atomsem)

Estrup and Anderson18 using the retarding potential technique and

Plummer and Bel120 using a field emission Fowler-Nordheim plot45 found a

work function change versus coverage relationship which deviated only slightly

8from being linear in near agreement with the results of Madey and Yates

46 47Becker using field emission and Armstrong using a retarding potential

technique found a work function change for the H W(100) system of approximately2

48090 eV when e = 10 Hopkins and Pender obtained an anomalously low value

49of 054 eV while Collins and Blott obtained a high value of 109 eV for the

same coverage

127 Chemical Probes

Hydrogen adsorption has also been studied using a chemical probe

ie by coadsorbing some additional different gas with the hydrogen and

observing the effect this additional species had on results obtained using

techniques discussed in previous sections

50Yates and Madey found using flash desorption that CO coadsorption

onto a hydrogen monolayer displaces both the Sl and S2 binding states with

near unity probability in doing so initially replacing these states with a

15

less tightly bound hydrogen state After high CO exposure no hydrogen remains

on the surface if the coadsorption is done at room temperature At lower

temperatures only about 10 percent of the hydrogen is desorbed Jelend and

Menzel lO 5l found that the ionic and total desorption cross sections for

hydrogen the sticking probability and bond strength of hydrogen are all

affected by preadsorption of CO or oxygen Yates and Madey3 found that N2

slowly replaces preadsorbed hydrogen at temperatures greater than 300oC Below

2730

K the displacement of hydrogen by N2 does not occur

These results all show that small impurity concentrations of CO N

and 0 if intentionally introduced do affect the resulting hydrogen chemishy

sorption properties These experiments underscore the necessity of surface

chemical characterization

128 Chemisorption Models

7 18 19 52 Several models have been proposed to account for the

experimental results from hydrogen chemisorption on W(lOO) Each of these

models embodies some combination of the following three phenomenological

assumptions 1) The tungsten surface is ideal planar and free of contamishy

nants 2) There are available two types of adsorbates namely hydrogen atoms

53and molecules 3) There are three high symmetry logical bonding sites on

a BCC (100) face usually designated by the letters A B and C These

positions are A a bridge site with 2 nearest neighbors and the hydrogen

equidistant from two adjacent surface tungsten atoms B a four fold symmetric

site with 5 nearest neighbors and the hydrogen located on a surface normal that

passes through the center of a surface primitive unit cell and C the hydrogen

16

located directly above a surfac~ atom with one nearest neighbor There are

15 152x10 A sitescm2 and 1x10 B ore sites~m2 on middotW(100)

The proposed models attempt to explain or at least be consistent

with empirical results presented in Sections 121 through 127 A summary

of salient features of the reported research includes a) a linear work function

15change with coverage b) a saturation coverage of between 15 and 20x10 H

2atomscm c) a coverage dependent adsorption layer symmetry including a we11shy

ordered surface at low coverages d) the appearance at low temperatures of a

surface translational symmetry not equivalent to room temperature clean

tungsten e) adsorption into states which produces a two state thermal desorpshy

tion spectrum with relative coverages in the two states of nQ n = 21 f) a ~1 ~2

coverage dependence for the observed electronic structure including the

absence at high coverage of structure observed at low coverages and g) no

observed vibrational levels which correspond to free H2 molecule

Following is a short review of several proposed surface models

18Estrup and Anderson account for their LEED spot pattern results

using a model wherein the hydrogen molecules dissociate upon adsorption and

migrate initially to form islands of hydrogen atoms located at type A sites in

a e(2x2) structure Dissociation is assumed since the e(2x2) pattern maximizes

15 2at a coverage of 25x10 mo1ecu1escm whereas without dissociation it would

maX1m1ze at 5 1015 mo ecu1es cm2bull e site adsorptionx 1 Although type B or type

could also produce the initial e(2x2) LEED pattern these sites are rejected

15 2because of the following 1) There are only 1x10 B or e sitescm This is

not enough sites to accommodate a full monolayer coverage if all binding is

atomic and if it is into equivalent sites as suggested by the linear work

17

function change with coverage 2) While a complete 14-order LEED pattern is

possible with type B or type C sites it is not possible to produce a 14-order

54pattern with (012) spots missing using B or C sites As the adsorption

14 2proceeds past a surface density of 5xlO hydrogen atomscm the adatoms conshy

tinuously rearrange into long period superlattices with antiphase domains

that produce successively the 12-order spot splitting the 13-order streak

(only partly ordered domains) the 14-order pattern and finally the (lxl)

full coverage structure

This model has been criticized for not accounting for the two

observed binding states and for predicting a maximum in intensity of the half

order spots at e = 025 these beams maximize at less than that coverage In

addition the authors themselves questioned the idea of well-ordered super-

lattices

In order to explain the relative intensity of the two states at

saturation seen in flash desorption n~~ = 21 Tamm and Schmidt7

propose a 1 2

model wherein the initial hydrogen adsorption is atomic and at B sites in a

C(2x2) pattern At full coverage the remaining B sites are filled with

hydrogen molecules B sites are chosen over C sites using molecular orbital

55arguments A sites are rejected as there are too many to satisfy this model

This model satisfies the second order kinetics observed for the ~2 state and

the first order kinetics observed for the ~l state In addition it explains

14 2 7 their saturation coverage of gt 3xl0 moleculescm One weakness of this

model pointed out by Tamm and Schmidt is that it is not obvious the model can

account for the observed LEED and work function results Also no molecular

vibrations have been observed

18

19yonehara and Schmidt propose a three bonding state model that is

more consistent with the low temperature LEED results In their model an

additional state labeled ~3 is always present at the surface in A or B

2sites with a density of 5xl014 atomscm bull This state orders on the surface at

low temperatures to give the C(2x2) pattern observed At room and higher

temperatures it goes into solution (or perhaps disorders) becoming invisible

to LEED In the type A site model adsorption from the background fills in the

remaining A sites sequentially with the ~2 and ~l states This model is similar

18to that proposed by Estrup and Anderson except that the ~3 state although

invisible to LEED at room and higher temperatures occupies 14 of the type A

sites This model eliminates the stoichiometric objection to Estrup and

Andersons model The model depends on the possibility of type A sites being

occupied by states with considerably different properties If the ~3 state

occupies type B sites ~2 state adsorption fills in the remaining type B sites

and then ~l hydrogen adsorbs into random or possibly type C sites This

second model does not explain satisfactorily work function or LEED spot pattern

experimental data

56Propst found using a point defect type calculation that only a

B-site model of adsorption would produce the observed vibrational energies of

5769 meV and 135 meV Park using LEED intensity arguments found that only

adsorption into sites with 4-fold symmetry on the (100) surface (type B or type

C sites) would be consistent with broadening and splitting of the half-order

beams while the integral beams remain sharp

8Madey and Yates offer a critique of several of the proposed models

v

33 Ion Species Identificationbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 74

34 Work Function Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 77

35 u+ Ion Desorption Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 83 351 ut Ion Energy Distributionbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 83 352 Desorption Curve Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 93 353 Temperature Dependence of the Fast State a+ Ion

Signal 4) 96

354 Additional Measurements on the Fast State bullbullbullbullbullbullbullbullbullbull 98

36 Low Energy Electron Diffraction Data bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull lll

37 Electron Scattering Data bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 119

4 DISCUSSION OF RESULTS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 126

41 Auger Electron Spectroscopy Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 126

42 Work Function Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 127

43 Electron Stimulated Desorption Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 129 431 Room Temperature Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 130 432 Low Temperature Fast State Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 135 433 Low Temperature Dosed Surface Results bullbullbullbullbullbullbullbullbullbullbullbullbull142

44 Low Energy Electron Diffraction Measurements bullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 145

45 Electron Scattering Results bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull 161

5 SUMMARY AND CONCLUSIONS bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull165

REFERENCES 171

APPENDIX A 0 176 poundI

APPENDIX B 9 c 179

APPENDIX C 0 182 0 bull

VITA 190lilt bullbullbullbullbullbullbullbullbullbullbullbullbullbullbullbull bullbull

1

CHAPTER 1

MOTIVATION AND REVIEW OF LITERATURE

11 Introduction and Motivation

The experimental investigation of chemisorption of gases onto metallic

surfaces has uncovered in the past decade a baffling complexity of adsorption

geometries and binding state kinetics and energetics A growing interest in

surface phenomena generated by the questions raised in these researches and

by the long anticipated use of the answers has led to the rapid development

of new surface probing techniques and to an increase in the availability of new

technologies compatible with surface research Aided by the new technologies

and techniques chemisorption research has been undertaken on many different

systems (system =solid substrate + gas adsorbate) under a wide range of

environmental conditions Surface theoreticians have proposed microscopic and

phenomenological models with which data have been analyzed and compared A

large experimental and theoretical literature in the field has resulted

The study of adsorption on metal surfaces has among its primary

objectives a detailed description of the physical geometry of the atoms in the

surface layers and a quantitative understanding of electronic and chemical

surface phenomena While considerable progress has been made in documenting

gas-surface phenomena it has been only in the past three years that any

credible claims to understanding specific surface overlayer geometries have

1been made At the same time many basic questions involving system kinetics

and energetics have not been adequately resolved 2

It has become apparent from the failure of the research effort to

satisfy the above objectives and from seemingly conflicting data that adequate

2

characterization is basic to all surface experimentation The need for such

characterization has been strongly emphasized by disagreement between quasishy

similar research performed at different locations Small impurity concentrashy

tions small changes in crystal orientation minor temperature differences etc

often have considerable effects on experimental measurements

A corollary to the necessity of being able to characterize the

surface is the importance of being able to control the surface characterization

In order to compare data it is necessary to reproduce all experimental paramshy

eters accurately The theory to which experimental data are compared is

constrained by complexity to be based on ideal clean substrates with a

definable gas-surface interaction It is desirable to be able to choose the

adsorbate-substrate system and experimental parameters that can also be used

with theoretical models Of course compromises are inevitable

One approach in surface research has been to incorporate several

experimental techniques into one apparatus At the present time this approach

is increasingly more common A combination of surface sensitive probes offers

many advantages It provides for more checks of the data against results

published in the literature It allows for better characterization of the

experimental parameters It allows surface phenomena to be probed in more than

one way in order to eliminate ambiguities andor to reinforce conclusions It

eliminates the uncertainties that arise from data taken at different locations

using the techniques separately hence it allows better correlations between

data taken using the different techniques

In this study several aspects of the chemisorption interaction of

hydrogen with the low index (100) plane of single crystal tungsten denoted

3

W(100) have been investigated The choice of the tungsten-hydrogen system is

motivated by several factors The literature raises questions on the unusual

state behavior observed when hydrogen adsorbs on W(100) questions which our

apparatus is especially well-suited to study In addition the interaction of

hydrogen with close packed planes of single crystal metals including W(100)

is of particular interest to theoreticians since hydrogen is the simplest

molecular gas as detailed in Sections 128 and 129 below recent theories

of chemisorption have been applied to the hydrogen-tungsten system Tungsten

was chosen as the substrate metal because of its relative ease of preparation

and cleaning and because the large body of literature on the hydrogen-tungsten

system indicates wide interest in this interaction

In particular this research is concentrated in three areas Tung~

sten when cleaned and subsequently cooled to below room temperature exhibits

an anomalous centered (2x2) denoted C(2x2) low energy electron diffraction

pattern The possibility that this is caused by hydrogen diffusing from the

bulk or adsorbing from the background has been investigated Secondly the

room temperature and low temperature hydrogen adsorption binding states have

been studied using electron stimulated desorption and electron spectroscopy

Finally low energy electron diffraction has been used in an investigation of

the atomic geometry of the clean tungsten surface and the geometry of one of

the hydrogen binding states

The remaining sections of this chapter summarize literature relating

to this research Chapter 2 details the construction and performance of the

experimental apparatus A presentation of the experimental data is made in

Chapter 3 A discussion of the results is made in Chapter 4 A summaryof

4

the research and conclusions from it are given in Chapter 5

12 Review of Literature

121 Introduction

Experimental investigation of the adsorption of hydrogen on tungsten

has been carried out using field emitter tips macroscopic polycrystalline

surfaces and all the low index orientation single crystal faces including

W(lOO) From the wide range of results it is clear that the substrate exerts

a strong influence on the adsorption properties The question has been raised

whether adsorption models postulated from polycrystalline substrates can be

3extended even qualitatively to single crystal systems Likewise there are

such sharp differences in the chemisorption behavior of hydrogen onto the

different single crystal tungsten surfaces that the specific physical process

governing adsorption on each face appears to be unique to that face This

review will therefore be concerned for the most part with literature dealing

with hydrogen adsorption on W(lOO) the gas-substrate system used in this

research

The following data review is presented in subsections each concenshy

trating on a specific experimental technique or class of experiments Subsecshy

tion 127 includes some gas mixing literature which was considered relevant

and Subsection 129 gives a bibliography of some related chemisorption

theoretical papers

122 Thermal Desorption

One technique widely employed in investigation of the kinetics of gas

456adsorption has been thermal or flash desorption By monitoring the system

5

pressure changes of a species while rapidly heating the crystal inferences

can be made on the existence of binding states adsorption energies and reaction

7 8 order Heating is accomplished by a focussed light beam electron bombardshy

7 9ment or ohmic losses bull

The strongest indication that hydrogen exhibits multiple state

adsorption behavior on W(lOO) comes from flash desorption data Tamm and

8Schmidt 7 and Madey and Yates found for room temperature adsorption that two

large peaks are observed in the plot of pressure of H2 versus temperature

These peaks are centered at 4500 K and approximately 550oK Only hydrogen

molecules are observed to be desorbed from the surface Using desorption rate

456theory the lower temperature (~l) state is found to obey first order

kinetics the peak temperature being independent of initial coverage The

higher temperature (~2) state obeys second order kinetics the peak temperature

moving down from approximately 5800 K to 5500 K as the initial coverage is

increased

In addition to the two main peaks several small peaks comprising a

total of less than 10 percent of the desorbed hydrogen are observed during a

flash when the crystal is cooled to nitrogen temperature before hydrogen

exposure These states are believed to result from adsorption onto edges and

other non-(lOO) crystal orientations 8

The dependence of the desorption energy Ed on coverage is weak

~l 8 ~2 8with Ed = 252 kcallmole and Ed = 323 kcalmole As would be predicted

from this result the ~2 state is largely populated before the ~l state begins

to fill The ratio of the populations in the states at full coverage n~1n~2

8 lOis 2111 Jelend and Menzel point out that the cross section for electron

6

desorption (see Subsection 125) from the ~1 state is considerably smaller

than that of the ~2 state a result not predicted from the experimental flash

desorption binding energies

11Tibbetts using a stepped temperature flash desorption technique

has shown that hydrogen is bound to the surface of po1ycrysta11ine tungsten

W(110) and W(lll) with a distribution of binding energies ranging from

162 kca1mo1e (07 eV) to 415 kca1mo1e (lS eV) Using this technique

desorption at a constant temperature from a distribution of states can be easily

distinguished from desorption from a state with a unique binding energy While

he did not extend his work to W(100) these results suggest that hydrogen does

not bind to the surface at a few energies but rather with a range of binding

energies Tibbetts also shows that the determination of the order of the

desorption process is difficult for adsorption into a range of binding energies

In experiments using a combination of hydrogen and deuterium as room

temperature adsorbates the elemental composition of the desorbed ~1 and ~2

states shows complete isotopic mixing between the states7S independent of the

adsorption sequence or ratios However for low temperature adsorption

complete state conversion is not observed for all the additional small peaks

Where mixing is observed atomic adsorption is suggested Where mixing is not

observed molecular adsorption is inferred

Some theoretical models indicate that multiple peak flash desorption

spectra may be due to binding energy changes caused by adsorbate-adsorbate

interactions12 or order-disorder transactions during f1ash3 13 rather than

actual distinct multiple binding states under the adsorption conditions

7

123 LEED and HEED

Low energy electron diffraction or LEED refers to the coherent

scattering of a monoenergetic focused electron beam from the surface of a

crystalline solid Electrons in the energy range 0 lt E lt 300 eV interact

strongly with the ion cores and valence electrons of the solid The inelastic

mean free path is on the order of 10 X As a consequence electrons which are

observed elastically have been backscattered within the first several surface

layers in the solid making LEED a surface sensitive probe 14 15 The periodic

translational symmetry of the surface leads to the momentum conservation laws

I

= (11)~I ~II + g

= (2mEh2)t sin reg (12)~II

where ~II is the momentum parallel to the surface of the emerging electron and

~ is the parallel momentum of the incident electron The vector g is a

reciprocal lattice vector of the solid surface and reg is the polar angle of the

incident beam As a result of the conservation laws the scattered electrons

emerge in a series of beams called a spot pattern which is a manifestation

of the surface symmetry The beams are labelled by their corresponding g

vectors Information on the actual configuration of the surface ion cores is

extracted from the plot of elastically scattered electron intensity versus

16 17incident energy called elastic intensity profiles or I-V curves

Changes in the LEED spot pattern resulting from hydrogen adsorption

18onto W(lOO) at room temperature have been reported by Estrup and Anderson and

19by Yonehara and Schmidt As hydrogen is adsorbed at room temperature the

8

(lxl) pattern characteristic of clean tungsten changes into a weak C(2x2)

pattern at a hydrogen coverage e of less than 005 monolayer (where e = 1

15 2monolayer at a coverage of 2xlO hydrogen atomsem which is 2 hydrogen atoms

for each surface tungsten atom) The (1212) spots become sharp and intense

with increased hydrogen adsorption reaching a maxima in intensity at approxishy

19 18mately 015 monolayer or e = 025 monolayer (Coverage in both cases was

determined by flash desorption) The (1212) spots split into four new spots

which move apart continuously and upon further hydrogen adsorption eventually

19form 13-order streaks at a coverage of ~ 035 monolayer These in turn

change at higher coverage into poorly developed 14-order features with the

(012) spots missing Finally near e = 10 all extra diffraction features

are gone and the LEED pattern returns to a (lxl) symmetry This series of LEED

pattern changes is reversible with coverage

For hydrogen adsorption at 1950 K the C(2x2) pattern develops

maximizes in intensity at e ~ 18 and splits into four beams similar to room

temperature adsorption However the higher coverage LEED structures third

19 0order streaks and fourth order spots are not observed At 78 K adsorption

of hydrogen produces half-order C(2x2) spots that maximize in intensity at

e ~ 027 Additional hydrogen adsorption beyond this causes the half-order

19spot intensity to decrease but with no splitting of the spots

If clean W(lOO) is allowed to cool below room temperature a sharp

19and intense C(2x2) pattern is observed even though total coverage due to the

background of H2 and CO is less than e = 001 as determined by flash desorption

measurements The intensity of these half-order beams can be altered by

additional hydrogen adsorption or by various hydrogen and oxygen heat treatmenm

9

l9Yonehara and Schmidt have postulated that the extra spots may result from

hydrogen endothermally soluble in tungsten diffusing to the surface Possible

alternative explanations include the reconstruction of the clean tungsten

surface or impurity over layer adsorption from the background It should be

20noted that Plummer and Bell saw no effects upon cooling their field emitter

tip to 78 0 K that they could attribute to hydrogen diffusion to the surface

Elastic intensity profiles for several beams have been measured by

18 21Estrup and Anderson and by Papageorgopoulos and Chen at room temperature

Estrup and Anderson reported profiles for (00) and (1212) tEED lJeams out

to 300 eV The intensity variation for the (OO) beam was found to be indepenshy

dent of hydrogen coverage a manifestation of the relatively weak scattering

cross section of a hydrogen ion core relative to that of tungsten The

additional beams observed on adsorption would be expected to be more sensitive

13to the adsorbed overlayer Estrup found the intensity of the (1212) beams

for H W(100) and D W(100) to decrease as a function of temperature between2 2

300o

K and 6000 K at a constant incident energy but neither obeyed the Debyeshy

22Waller relation Estrup interpreted this result to imply the adsorbed

species are directly involved in the diffraction process

19Yonehara examined the tEED spot intensities as a function of

temperature down to liquid nitrogen temperatures His results are examined

in the data discussion in Chapter 4

The hydrogen-WlOO) system has also been studied using high energy

electron diffraction or HEED23 This technique utilizes the strong forward

scattering of high energy electrons in probing the surface by having the

incident beam strike the solid at grazing incidence The resulting diffraction

10

patterns are sensitive to the translational symmetry of the surface As in

LEED a diffraction pattern with C(2x2) symmetry is observed during initial

hydrogen adsorption With further hydrogen exposure the beams broaden some

splitting is observed and eventually the half-order intensity decreases

completely The width of the observed half-order beams has been interpreted to

indicate hydrogen initially adsorbs into a large number of small square or

rectangular domains each consisting of 20 to 25 atoms with sides parallel to

the [100J directions

124 Electron Scattering Field Emission and Photoemission

The electronic structure of the surface layers of solids can be

characterized by electron emission spectroscopies Such techniques require a

surface sensitive perturbation on the gas-solid system leading to an electron

yield A spectrometer is used to energy analyze the emitted or scattered

24 25electrons In electron scatter~ng a monoenergetic beam of electrons is

directed onto a crystal The inelastically scattered signal containing

quantitative information about electronic excitations at the surface is

measured as a function of energy loss Field emission is a technique in which

application of a high positive electric field to the surface lowers the potenshy

26tia1 barrier for electrons to tunnel out of the system The total energy

distribution of electrons emitted as a result of the applied field contains

bull 27 28spectroscopic data on the adsorbed spec~es Structure induced by the

adsorbate in the elastic tunneling spectrum is thought to be a measure of the

d f i h f h d b 2930e1ectron~c ens~ty 0 states n t e v~c~n~ty 0 tea sor ate The

inelastic tunneling spectrum reveals discrete excitation energies in the

11

gas-surface system In photoemission or ftotoelectron spectroscopy an incident

monochromatic light beam incident on the solid induces emission of electrons by

interband transitions with or without an inelastic scattering after the initial

31 exc i tat~on

The vibrational spectrum of hydrogen adsorbed on the tungsten (100)

32surface has been observed by Propst and Piper using inelastic low energy

20electron scattering and by Plummer and Bell using field emission Propst

and Piper found two peaks in the energy loss distribution E - E pr~mary scattered

at 135 meV and 69 meV for all coverages used up to saturation These results

are interpreted as indicating hydrogen atoms are multiply bonded to tungsten

33atoms No losses near 550 meV the vibrational energy of molecular hydrogen

were observed Plummer and Bell measured the total energy distribution of

field emitted electrons from tungsten (100) as a function of coverage of

hydrogen and deuterium at several temperatures Discrete losses at 100 meV

and 50 meV for deuterium which correspond to the 135 meV and 69 meV vibrational

losses observed by Propst and Piper were evident for deuterium in the a state2

No discrete energy losses at 550 meV or 390 meV the molecular vibrational

levels of H2 and D2 respectively were measured Hence there was no observable

molecular adsorption at room temperature

Additional structure in the field emission total energy distribution

is observed as a function of hydrogen adsorption The clean surface spectrum

has two peaks 04 eV and 15 eV down from the Fermi level which have been

d ifi d h h d f d h h f 34 A h d ~ ent e as ~g ens~ty 0 states assoc~ate w~t t e sur ace s y rogen

(or deuterium) coverage increases between 0 and 5xl014 atomscm2

ie while

the ~2 state is filling the surface states disappear and the field emitted

12

energy distribution changes from that of clean tungsten to a spectrum indicative

of resonance tunneling through a level positioned 09 eV below the Fermi energy

There is a shoulder at -11 eV and an energy dependent background With

further adsorption up to saturation this energy distribution characteristic

of the ~2 state shifts to a broad level centered at 26 eV below the Fermi

level This is interpreted to indicate that the ~1 state does not adsorb on

top of the ~2 state

35The photoemission results of Wac1awski and P1ummer and Feuerbacher

36and Fitton are in agreement with those obtained using the field emission

technique As hydrogen adsorbs on the surface emission from the 04 eV

surface state decreases The level 09 eV down from the Fermi energy as well

as a level 5 eV down remain unchanged by adsorption At full coverage a strong

peak at -25 eV is observed

125 Electron Stimulated Desorption

Desorption of energetic ions and neutrals from a gas over layer can

be induced by electron bombardment This process termed electron stimulated

desorption or ESD37383940 is described below in the discussion of the data

in Chapter 4 In brief intensity and energy measurements of the ion yield can

be interpreted in terms of the parameters characterizing the binding energy

curve of the adsorbed state and adsorption states can be classified by cross

section for desorption_ ion energy distributions primary energy threshold for

desorption etc

ESD has been employed to study adsorption kinetics for the hydrogenshy

41 ~ tungsten (100) system Madey obtained a coverage dependence yield of H ions

13

observing an ion signal during the initial stages of hydrogen adsorption that

reached a maximum at a coverage of e = 017 and then decayed to zero yield

23 2after e = 075 the maximum cross section obtained was 18x10- cm at a

primary electron energy of 100 eV Tamm and Schmidt7 reported a maximum cross

section for H+ desorption of 10- 21 cm 2bull Benninghoven Loebach and Treitz42

10and Je1end and Menze1 obtained qualitatively similar (to Madey) coverage

dependencies on tungsten ribbons which were believed to be predominantly of

(100) crystallites the main difference between these results and Madeys is

the observation of a residual signal at high hydrogen coverage

The cross section results indicate that the a state even though it2

is more tightly bound than the a state provides more hydrogen ions upon1

electron impact The cross section for ion desorption from the a1 state is

d41 10-25est1mate to be 1ess t han cm2 bull

+No energy distributions of the desorbed H ions have been reported

The dependence of the desorption cross section on the primary electron energy

10has not been investigated Je1end and Menzel found their results independent

of temperature for temperatures above room temperature No ESD results for

otemperatures less than 300 K have been reported

126 Work Function

The work function is the minimum amount of energy required to remove

43 an electron from a meta1 An adsorbate may alter the potential barrier for

electrons at the surface and hence change the work function Such is the case

for hydrogen on W(100) Hydrogen adsorbs as an electronegative ion on W(100)

thereby increasing the work function

14

44Using a retarding potential current-voltage techn~que Madey and

8Yates reported a linear relation between work function change and hydrogen

coverage over the entire coverage range for adsorption occurring at a

temperature 0 f 2= 300 0 K Coverages were determined by flash desorption

Total work function change at one monolayer was found to be 090 eV which

corresponds to a constant dipole moment per adsorbed atom of 015 Debye

15 2(assuming a total coverage of 15xlO atomsem)

Estrup and Anderson18 using the retarding potential technique and

Plummer and Bel120 using a field emission Fowler-Nordheim plot45 found a

work function change versus coverage relationship which deviated only slightly

8from being linear in near agreement with the results of Madey and Yates

46 47Becker using field emission and Armstrong using a retarding potential

technique found a work function change for the H W(100) system of approximately2

48090 eV when e = 10 Hopkins and Pender obtained an anomalously low value

49of 054 eV while Collins and Blott obtained a high value of 109 eV for the

same coverage

127 Chemical Probes

Hydrogen adsorption has also been studied using a chemical probe

ie by coadsorbing some additional different gas with the hydrogen and

observing the effect this additional species had on results obtained using

techniques discussed in previous sections

50Yates and Madey found using flash desorption that CO coadsorption

onto a hydrogen monolayer displaces both the Sl and S2 binding states with

near unity probability in doing so initially replacing these states with a

15

less tightly bound hydrogen state After high CO exposure no hydrogen remains

on the surface if the coadsorption is done at room temperature At lower

temperatures only about 10 percent of the hydrogen is desorbed Jelend and

Menzel lO 5l found that the ionic and total desorption cross sections for

hydrogen the sticking probability and bond strength of hydrogen are all

affected by preadsorption of CO or oxygen Yates and Madey3 found that N2

slowly replaces preadsorbed hydrogen at temperatures greater than 300oC Below

2730

K the displacement of hydrogen by N2 does not occur

These results all show that small impurity concentrations of CO N

and 0 if intentionally introduced do affect the resulting hydrogen chemishy

sorption properties These experiments underscore the necessity of surface

chemical characterization

128 Chemisorption Models

7 18 19 52 Several models have been proposed to account for the

experimental results from hydrogen chemisorption on W(lOO) Each of these

models embodies some combination of the following three phenomenological

assumptions 1) The tungsten surface is ideal planar and free of contamishy

nants 2) There are available two types of adsorbates namely hydrogen atoms

53and molecules 3) There are three high symmetry logical bonding sites on

a BCC (100) face usually designated by the letters A B and C These

positions are A a bridge site with 2 nearest neighbors and the hydrogen

equidistant from two adjacent surface tungsten atoms B a four fold symmetric

site with 5 nearest neighbors and the hydrogen located on a surface normal that

passes through the center of a surface primitive unit cell and C the hydrogen

16

located directly above a surfac~ atom with one nearest neighbor There are

15 152x10 A sitescm2 and 1x10 B ore sites~m2 on middotW(100)

The proposed models attempt to explain or at least be consistent

with empirical results presented in Sections 121 through 127 A summary

of salient features of the reported research includes a) a linear work function

15change with coverage b) a saturation coverage of between 15 and 20x10 H

2atomscm c) a coverage dependent adsorption layer symmetry including a we11shy

ordered surface at low coverages d) the appearance at low temperatures of a

surface translational symmetry not equivalent to room temperature clean

tungsten e) adsorption into states which produces a two state thermal desorpshy

tion spectrum with relative coverages in the two states of nQ n = 21 f) a ~1 ~2

coverage dependence for the observed electronic structure including the

absence at high coverage of structure observed at low coverages and g) no

observed vibrational levels which correspond to free H2 molecule

Following is a short review of several proposed surface models

18Estrup and Anderson account for their LEED spot pattern results

using a model wherein the hydrogen molecules dissociate upon adsorption and

migrate initially to form islands of hydrogen atoms located at type A sites in

a e(2x2) structure Dissociation is assumed since the e(2x2) pattern maximizes

15 2at a coverage of 25x10 mo1ecu1escm whereas without dissociation it would

maX1m1ze at 5 1015 mo ecu1es cm2bull e site adsorptionx 1 Although type B or type

could also produce the initial e(2x2) LEED pattern these sites are rejected

15 2because of the following 1) There are only 1x10 B or e sitescm This is

not enough sites to accommodate a full monolayer coverage if all binding is

atomic and if it is into equivalent sites as suggested by the linear work

17

function change with coverage 2) While a complete 14-order LEED pattern is

possible with type B or type C sites it is not possible to produce a 14-order

54pattern with (012) spots missing using B or C sites As the adsorption

14 2proceeds past a surface density of 5xlO hydrogen atomscm the adatoms conshy

tinuously rearrange into long period superlattices with antiphase domains

that produce successively the 12-order spot splitting the 13-order streak

(only partly ordered domains) the 14-order pattern and finally the (lxl)

full coverage structure

This model has been criticized for not accounting for the two

observed binding states and for predicting a maximum in intensity of the half

order spots at e = 025 these beams maximize at less than that coverage In

addition the authors themselves questioned the idea of well-ordered super-

lattices

In order to explain the relative intensity of the two states at

saturation seen in flash desorption n~~ = 21 Tamm and Schmidt7

propose a 1 2

model wherein the initial hydrogen adsorption is atomic and at B sites in a

C(2x2) pattern At full coverage the remaining B sites are filled with

hydrogen molecules B sites are chosen over C sites using molecular orbital

55arguments A sites are rejected as there are too many to satisfy this model

This model satisfies the second order kinetics observed for the ~2 state and

the first order kinetics observed for the ~l state In addition it explains

14 2 7 their saturation coverage of gt 3xl0 moleculescm One weakness of this

model pointed out by Tamm and Schmidt is that it is not obvious the model can

account for the observed LEED and work function results Also no molecular

vibrations have been observed

18

19yonehara and Schmidt propose a three bonding state model that is

more consistent with the low temperature LEED results In their model an

additional state labeled ~3 is always present at the surface in A or B

2sites with a density of 5xl014 atomscm bull This state orders on the surface at

low temperatures to give the C(2x2) pattern observed At room and higher

temperatures it goes into solution (or perhaps disorders) becoming invisible

to LEED In the type A site model adsorption from the background fills in the

remaining A sites sequentially with the ~2 and ~l states This model is similar

18to that proposed by Estrup and Anderson except that the ~3 state although

invisible to LEED at room and higher temperatures occupies 14 of the type A

sites This model eliminates the stoichiometric objection to Estrup and

Andersons model The model depends on the possibility of type A sites being

occupied by states with considerably different properties If the ~3 state

occupies type B sites ~2 state adsorption fills in the remaining type B sites

and then ~l hydrogen adsorbs into random or possibly type C sites This

second model does not explain satisfactorily work function or LEED spot pattern

experimental data

56Propst found using a point defect type calculation that only a

B-site model of adsorption would produce the observed vibrational energies of

5769 meV and 135 meV Park using LEED intensity arguments found that only

adsorption into sites with 4-fold symmetry on the (100) surface (type B or type

C sites) would be consistent with broadening and splitting of the half-order

beams while the integral beams remain sharp

8Madey and Yates offer a critique of several of the proposed models

1

CHAPTER 1

MOTIVATION AND REVIEW OF LITERATURE

11 Introduction and Motivation

The experimental investigation of chemisorption of gases onto metallic

surfaces has uncovered in the past decade a baffling complexity of adsorption

geometries and binding state kinetics and energetics A growing interest in

surface phenomena generated by the questions raised in these researches and

by the long anticipated use of the answers has led to the rapid development

of new surface probing techniques and to an increase in the availability of new

technologies compatible with surface research Aided by the new technologies

and techniques chemisorption research has been undertaken on many different

systems (system =solid substrate + gas adsorbate) under a wide range of

environmental conditions Surface theoreticians have proposed microscopic and

phenomenological models with which data have been analyzed and compared A

large experimental and theoretical literature in the field has resulted

The study of adsorption on metal surfaces has among its primary

objectives a detailed description of the physical geometry of the atoms in the

surface layers and a quantitative understanding of electronic and chemical

surface phenomena While considerable progress has been made in documenting

gas-surface phenomena it has been only in the past three years that any

credible claims to understanding specific surface overlayer geometries have

1been made At the same time many basic questions involving system kinetics

and energetics have not been adequately resolved 2

It has become apparent from the failure of the research effort to

satisfy the above objectives and from seemingly conflicting data that adequate

2

characterization is basic to all surface experimentation The need for such

characterization has been strongly emphasized by disagreement between quasishy

similar research performed at different locations Small impurity concentrashy

tions small changes in crystal orientation minor temperature differences etc

often have considerable effects on experimental measurements

A corollary to the necessity of being able to characterize the

surface is the importance of being able to control the surface characterization

In order to compare data it is necessary to reproduce all experimental paramshy

eters accurately The theory to which experimental data are compared is

constrained by complexity to be based on ideal clean substrates with a

definable gas-surface interaction It is desirable to be able to choose the

adsorbate-substrate system and experimental parameters that can also be used

with theoretical models Of course compromises are inevitable

One approach in surface research has been to incorporate several

experimental techniques into one apparatus At the present time this approach

is increasingly more common A combination of surface sensitive probes offers

many advantages It provides for more checks of the data against results

published in the literature It allows for better characterization of the

experimental parameters It allows surface phenomena to be probed in more than

one way in order to eliminate ambiguities andor to reinforce conclusions It

eliminates the uncertainties that arise from data taken at different locations

using the techniques separately hence it allows better correlations between

data taken using the different techniques

In this study several aspects of the chemisorption interaction of

hydrogen with the low index (100) plane of single crystal tungsten denoted

3

W(100) have been investigated The choice of the tungsten-hydrogen system is

motivated by several factors The literature raises questions on the unusual

state behavior observed when hydrogen adsorbs on W(100) questions which our

apparatus is especially well-suited to study In addition the interaction of

hydrogen with close packed planes of single crystal metals including W(100)

is of particular interest to theoreticians since hydrogen is the simplest

molecular gas as detailed in Sections 128 and 129 below recent theories

of chemisorption have been applied to the hydrogen-tungsten system Tungsten

was chosen as the substrate metal because of its relative ease of preparation

and cleaning and because the large body of literature on the hydrogen-tungsten

system indicates wide interest in this interaction

In particular this research is concentrated in three areas Tung~

sten when cleaned and subsequently cooled to below room temperature exhibits

an anomalous centered (2x2) denoted C(2x2) low energy electron diffraction

pattern The possibility that this is caused by hydrogen diffusing from the

bulk or adsorbing from the background has been investigated Secondly the

room temperature and low temperature hydrogen adsorption binding states have

been studied using electron stimulated desorption and electron spectroscopy

Finally low energy electron diffraction has been used in an investigation of

the atomic geometry of the clean tungsten surface and the geometry of one of

the hydrogen binding states

The remaining sections of this chapter summarize literature relating

to this research Chapter 2 details the construction and performance of the

experimental apparatus A presentation of the experimental data is made in

Chapter 3 A discussion of the results is made in Chapter 4 A summaryof

4

the research and conclusions from it are given in Chapter 5

12 Review of Literature

121 Introduction

Experimental investigation of the adsorption of hydrogen on tungsten

has been carried out using field emitter tips macroscopic polycrystalline

surfaces and all the low index orientation single crystal faces including

W(lOO) From the wide range of results it is clear that the substrate exerts

a strong influence on the adsorption properties The question has been raised

whether adsorption models postulated from polycrystalline substrates can be

3extended even qualitatively to single crystal systems Likewise there are

such sharp differences in the chemisorption behavior of hydrogen onto the

different single crystal tungsten surfaces that the specific physical process

governing adsorption on each face appears to be unique to that face This

review will therefore be concerned for the most part with literature dealing

with hydrogen adsorption on W(lOO) the gas-substrate system used in this

research

The following data review is presented in subsections each concenshy

trating on a specific experimental technique or class of experiments Subsecshy

tion 127 includes some gas mixing literature which was considered relevant

and Subsection 129 gives a bibliography of some related chemisorption

theoretical papers

122 Thermal Desorption

One technique widely employed in investigation of the kinetics of gas

456adsorption has been thermal or flash desorption By monitoring the system

5

pressure changes of a species while rapidly heating the crystal inferences

can be made on the existence of binding states adsorption energies and reaction

7 8 order Heating is accomplished by a focussed light beam electron bombardshy

7 9ment or ohmic losses bull

The strongest indication that hydrogen exhibits multiple state

adsorption behavior on W(lOO) comes from flash desorption data Tamm and

8Schmidt 7 and Madey and Yates found for room temperature adsorption that two

large peaks are observed in the plot of pressure of H2 versus temperature

These peaks are centered at 4500 K and approximately 550oK Only hydrogen

molecules are observed to be desorbed from the surface Using desorption rate

456theory the lower temperature (~l) state is found to obey first order

kinetics the peak temperature being independent of initial coverage The

higher temperature (~2) state obeys second order kinetics the peak temperature

moving down from approximately 5800 K to 5500 K as the initial coverage is

increased

In addition to the two main peaks several small peaks comprising a

total of less than 10 percent of the desorbed hydrogen are observed during a

flash when the crystal is cooled to nitrogen temperature before hydrogen

exposure These states are believed to result from adsorption onto edges and

other non-(lOO) crystal orientations 8

The dependence of the desorption energy Ed on coverage is weak

~l 8 ~2 8with Ed = 252 kcallmole and Ed = 323 kcalmole As would be predicted

from this result the ~2 state is largely populated before the ~l state begins

to fill The ratio of the populations in the states at full coverage n~1n~2

8 lOis 2111 Jelend and Menzel point out that the cross section for electron

6

desorption (see Subsection 125) from the ~1 state is considerably smaller

than that of the ~2 state a result not predicted from the experimental flash

desorption binding energies

11Tibbetts using a stepped temperature flash desorption technique

has shown that hydrogen is bound to the surface of po1ycrysta11ine tungsten

W(110) and W(lll) with a distribution of binding energies ranging from

162 kca1mo1e (07 eV) to 415 kca1mo1e (lS eV) Using this technique

desorption at a constant temperature from a distribution of states can be easily

distinguished from desorption from a state with a unique binding energy While

he did not extend his work to W(100) these results suggest that hydrogen does

not bind to the surface at a few energies but rather with a range of binding

energies Tibbetts also shows that the determination of the order of the

desorption process is difficult for adsorption into a range of binding energies

In experiments using a combination of hydrogen and deuterium as room

temperature adsorbates the elemental composition of the desorbed ~1 and ~2

states shows complete isotopic mixing between the states7S independent of the

adsorption sequence or ratios However for low temperature adsorption

complete state conversion is not observed for all the additional small peaks

Where mixing is observed atomic adsorption is suggested Where mixing is not

observed molecular adsorption is inferred

Some theoretical models indicate that multiple peak flash desorption

spectra may be due to binding energy changes caused by adsorbate-adsorbate

interactions12 or order-disorder transactions during f1ash3 13 rather than

actual distinct multiple binding states under the adsorption conditions

7

123 LEED and HEED

Low energy electron diffraction or LEED refers to the coherent

scattering of a monoenergetic focused electron beam from the surface of a

crystalline solid Electrons in the energy range 0 lt E lt 300 eV interact

strongly with the ion cores and valence electrons of the solid The inelastic

mean free path is on the order of 10 X As a consequence electrons which are

observed elastically have been backscattered within the first several surface

layers in the solid making LEED a surface sensitive probe 14 15 The periodic

translational symmetry of the surface leads to the momentum conservation laws

I

= (11)~I ~II + g

= (2mEh2)t sin reg (12)~II

where ~II is the momentum parallel to the surface of the emerging electron and

~ is the parallel momentum of the incident electron The vector g is a

reciprocal lattice vector of the solid surface and reg is the polar angle of the

incident beam As a result of the conservation laws the scattered electrons

emerge in a series of beams called a spot pattern which is a manifestation

of the surface symmetry The beams are labelled by their corresponding g

vectors Information on the actual configuration of the surface ion cores is

extracted from the plot of elastically scattered electron intensity versus

16 17incident energy called elastic intensity profiles or I-V curves

Changes in the LEED spot pattern resulting from hydrogen adsorption

18onto W(lOO) at room temperature have been reported by Estrup and Anderson and

19by Yonehara and Schmidt As hydrogen is adsorbed at room temperature the

8

(lxl) pattern characteristic of clean tungsten changes into a weak C(2x2)

pattern at a hydrogen coverage e of less than 005 monolayer (where e = 1

15 2monolayer at a coverage of 2xlO hydrogen atomsem which is 2 hydrogen atoms

for each surface tungsten atom) The (1212) spots become sharp and intense

with increased hydrogen adsorption reaching a maxima in intensity at approxishy

19 18mately 015 monolayer or e = 025 monolayer (Coverage in both cases was

determined by flash desorption) The (1212) spots split into four new spots

which move apart continuously and upon further hydrogen adsorption eventually

19form 13-order streaks at a coverage of ~ 035 monolayer These in turn

change at higher coverage into poorly developed 14-order features with the

(012) spots missing Finally near e = 10 all extra diffraction features

are gone and the LEED pattern returns to a (lxl) symmetry This series of LEED

pattern changes is reversible with coverage

For hydrogen adsorption at 1950 K the C(2x2) pattern develops

maximizes in intensity at e ~ 18 and splits into four beams similar to room

temperature adsorption However the higher coverage LEED structures third

19 0order streaks and fourth order spots are not observed At 78 K adsorption

of hydrogen produces half-order C(2x2) spots that maximize in intensity at

e ~ 027 Additional hydrogen adsorption beyond this causes the half-order

19spot intensity to decrease but with no splitting of the spots

If clean W(lOO) is allowed to cool below room temperature a sharp

19and intense C(2x2) pattern is observed even though total coverage due to the

background of H2 and CO is less than e = 001 as determined by flash desorption

measurements The intensity of these half-order beams can be altered by

additional hydrogen adsorption or by various hydrogen and oxygen heat treatmenm

9

l9Yonehara and Schmidt have postulated that the extra spots may result from

hydrogen endothermally soluble in tungsten diffusing to the surface Possible

alternative explanations include the reconstruction of the clean tungsten

surface or impurity over layer adsorption from the background It should be

20noted that Plummer and Bell saw no effects upon cooling their field emitter

tip to 78 0 K that they could attribute to hydrogen diffusion to the surface

Elastic intensity profiles for several beams have been measured by

18 21Estrup and Anderson and by Papageorgopoulos and Chen at room temperature

Estrup and Anderson reported profiles for (00) and (1212) tEED lJeams out

to 300 eV The intensity variation for the (OO) beam was found to be indepenshy

dent of hydrogen coverage a manifestation of the relatively weak scattering

cross section of a hydrogen ion core relative to that of tungsten The

additional beams observed on adsorption would be expected to be more sensitive

13to the adsorbed overlayer Estrup found the intensity of the (1212) beams

for H W(100) and D W(100) to decrease as a function of temperature between2 2

300o

K and 6000 K at a constant incident energy but neither obeyed the Debyeshy

22Waller relation Estrup interpreted this result to imply the adsorbed

species are directly involved in the diffraction process

19Yonehara examined the tEED spot intensities as a function of

temperature down to liquid nitrogen temperatures His results are examined

in the data discussion in Chapter 4

The hydrogen-WlOO) system has also been studied using high energy

electron diffraction or HEED23 This technique utilizes the strong forward

scattering of high energy electrons in probing the surface by having the

incident beam strike the solid at grazing incidence The resulting diffraction

10

patterns are sensitive to the translational symmetry of the surface As in

LEED a diffraction pattern with C(2x2) symmetry is observed during initial

hydrogen adsorption With further hydrogen exposure the beams broaden some

splitting is observed and eventually the half-order intensity decreases

completely The width of the observed half-order beams has been interpreted to

indicate hydrogen initially adsorbs into a large number of small square or

rectangular domains each consisting of 20 to 25 atoms with sides parallel to

the [100J directions

124 Electron Scattering Field Emission and Photoemission

The electronic structure of the surface layers of solids can be

characterized by electron emission spectroscopies Such techniques require a

surface sensitive perturbation on the gas-solid system leading to an electron

yield A spectrometer is used to energy analyze the emitted or scattered

24 25electrons In electron scatter~ng a monoenergetic beam of electrons is

directed onto a crystal The inelastically scattered signal containing

quantitative information about electronic excitations at the surface is

measured as a function of energy loss Field emission is a technique in which

application of a high positive electric field to the surface lowers the potenshy

26tia1 barrier for electrons to tunnel out of the system The total energy

distribution of electrons emitted as a result of the applied field contains

bull 27 28spectroscopic data on the adsorbed spec~es Structure induced by the

adsorbate in the elastic tunneling spectrum is thought to be a measure of the

d f i h f h d b 2930e1ectron~c ens~ty 0 states n t e v~c~n~ty 0 tea sor ate The

inelastic tunneling spectrum reveals discrete excitation energies in the

11

gas-surface system In photoemission or ftotoelectron spectroscopy an incident

monochromatic light beam incident on the solid induces emission of electrons by

interband transitions with or without an inelastic scattering after the initial

31 exc i tat~on

The vibrational spectrum of hydrogen adsorbed on the tungsten (100)

32surface has been observed by Propst and Piper using inelastic low energy

20electron scattering and by Plummer and Bell using field emission Propst

and Piper found two peaks in the energy loss distribution E - E pr~mary scattered

at 135 meV and 69 meV for all coverages used up to saturation These results

are interpreted as indicating hydrogen atoms are multiply bonded to tungsten

33atoms No losses near 550 meV the vibrational energy of molecular hydrogen

were observed Plummer and Bell measured the total energy distribution of

field emitted electrons from tungsten (100) as a function of coverage of

hydrogen and deuterium at several temperatures Discrete losses at 100 meV

and 50 meV for deuterium which correspond to the 135 meV and 69 meV vibrational

losses observed by Propst and Piper were evident for deuterium in the a state2

No discrete energy losses at 550 meV or 390 meV the molecular vibrational

levels of H2 and D2 respectively were measured Hence there was no observable

molecular adsorption at room temperature

Additional structure in the field emission total energy distribution

is observed as a function of hydrogen adsorption The clean surface spectrum

has two peaks 04 eV and 15 eV down from the Fermi level which have been

d ifi d h h d f d h h f 34 A h d ~ ent e as ~g ens~ty 0 states assoc~ate w~t t e sur ace s y rogen

(or deuterium) coverage increases between 0 and 5xl014 atomscm2

ie while

the ~2 state is filling the surface states disappear and the field emitted

12

energy distribution changes from that of clean tungsten to a spectrum indicative

of resonance tunneling through a level positioned 09 eV below the Fermi energy

There is a shoulder at -11 eV and an energy dependent background With

further adsorption up to saturation this energy distribution characteristic

of the ~2 state shifts to a broad level centered at 26 eV below the Fermi

level This is interpreted to indicate that the ~1 state does not adsorb on

top of the ~2 state

35The photoemission results of Wac1awski and P1ummer and Feuerbacher

36and Fitton are in agreement with those obtained using the field emission

technique As hydrogen adsorbs on the surface emission from the 04 eV

surface state decreases The level 09 eV down from the Fermi energy as well

as a level 5 eV down remain unchanged by adsorption At full coverage a strong

peak at -25 eV is observed

125 Electron Stimulated Desorption

Desorption of energetic ions and neutrals from a gas over layer can

be induced by electron bombardment This process termed electron stimulated

desorption or ESD37383940 is described below in the discussion of the data

in Chapter 4 In brief intensity and energy measurements of the ion yield can

be interpreted in terms of the parameters characterizing the binding energy

curve of the adsorbed state and adsorption states can be classified by cross

section for desorption_ ion energy distributions primary energy threshold for

desorption etc

ESD has been employed to study adsorption kinetics for the hydrogenshy

41 ~ tungsten (100) system Madey obtained a coverage dependence yield of H ions

13

observing an ion signal during the initial stages of hydrogen adsorption that

reached a maximum at a coverage of e = 017 and then decayed to zero yield

23 2after e = 075 the maximum cross section obtained was 18x10- cm at a

primary electron energy of 100 eV Tamm and Schmidt7 reported a maximum cross

section for H+ desorption of 10- 21 cm 2bull Benninghoven Loebach and Treitz42

10and Je1end and Menze1 obtained qualitatively similar (to Madey) coverage

dependencies on tungsten ribbons which were believed to be predominantly of

(100) crystallites the main difference between these results and Madeys is

the observation of a residual signal at high hydrogen coverage

The cross section results indicate that the a state even though it2

is more tightly bound than the a state provides more hydrogen ions upon1

electron impact The cross section for ion desorption from the a1 state is

d41 10-25est1mate to be 1ess t han cm2 bull

+No energy distributions of the desorbed H ions have been reported

The dependence of the desorption cross section on the primary electron energy

10has not been investigated Je1end and Menzel found their results independent

of temperature for temperatures above room temperature No ESD results for

otemperatures less than 300 K have been reported

126 Work Function

The work function is the minimum amount of energy required to remove

43 an electron from a meta1 An adsorbate may alter the potential barrier for

electrons at the surface and hence change the work function Such is the case

for hydrogen on W(100) Hydrogen adsorbs as an electronegative ion on W(100)

thereby increasing the work function

14

44Using a retarding potential current-voltage techn~que Madey and

8Yates reported a linear relation between work function change and hydrogen

coverage over the entire coverage range for adsorption occurring at a

temperature 0 f 2= 300 0 K Coverages were determined by flash desorption

Total work function change at one monolayer was found to be 090 eV which

corresponds to a constant dipole moment per adsorbed atom of 015 Debye

15 2(assuming a total coverage of 15xlO atomsem)

Estrup and Anderson18 using the retarding potential technique and

Plummer and Bel120 using a field emission Fowler-Nordheim plot45 found a

work function change versus coverage relationship which deviated only slightly

8from being linear in near agreement with the results of Madey and Yates

46 47Becker using field emission and Armstrong using a retarding potential

technique found a work function change for the H W(100) system of approximately2

48090 eV when e = 10 Hopkins and Pender obtained an anomalously low value

49of 054 eV while Collins and Blott obtained a high value of 109 eV for the

same coverage

127 Chemical Probes

Hydrogen adsorption has also been studied using a chemical probe

ie by coadsorbing some additional different gas with the hydrogen and

observing the effect this additional species had on results obtained using

techniques discussed in previous sections

50Yates and Madey found using flash desorption that CO coadsorption

onto a hydrogen monolayer displaces both the Sl and S2 binding states with

near unity probability in doing so initially replacing these states with a

15

less tightly bound hydrogen state After high CO exposure no hydrogen remains

on the surface if the coadsorption is done at room temperature At lower

temperatures only about 10 percent of the hydrogen is desorbed Jelend and

Menzel lO 5l found that the ionic and total desorption cross sections for

hydrogen the sticking probability and bond strength of hydrogen are all

affected by preadsorption of CO or oxygen Yates and Madey3 found that N2

slowly replaces preadsorbed hydrogen at temperatures greater than 300oC Below

2730

K the displacement of hydrogen by N2 does not occur

These results all show that small impurity concentrations of CO N

and 0 if intentionally introduced do affect the resulting hydrogen chemishy

sorption properties These experiments underscore the necessity of surface

chemical characterization

128 Chemisorption Models

7 18 19 52 Several models have been proposed to account for the

experimental results from hydrogen chemisorption on W(lOO) Each of these

models embodies some combination of the following three phenomenological

assumptions 1) The tungsten surface is ideal planar and free of contamishy

nants 2) There are available two types of adsorbates namely hydrogen atoms

53and molecules 3) There are three high symmetry logical bonding sites on

a BCC (100) face usually designated by the letters A B and C These

positions are A a bridge site with 2 nearest neighbors and the hydrogen

equidistant from two adjacent surface tungsten atoms B a four fold symmetric

site with 5 nearest neighbors and the hydrogen located on a surface normal that

passes through the center of a surface primitive unit cell and C the hydrogen

16

located directly above a surfac~ atom with one nearest neighbor There are

15 152x10 A sitescm2 and 1x10 B ore sites~m2 on middotW(100)

The proposed models attempt to explain or at least be consistent

with empirical results presented in Sections 121 through 127 A summary

of salient features of the reported research includes a) a linear work function

15change with coverage b) a saturation coverage of between 15 and 20x10 H

2atomscm c) a coverage dependent adsorption layer symmetry including a we11shy

ordered surface at low coverages d) the appearance at low temperatures of a

surface translational symmetry not equivalent to room temperature clean

tungsten e) adsorption into states which produces a two state thermal desorpshy

tion spectrum with relative coverages in the two states of nQ n = 21 f) a ~1 ~2

coverage dependence for the observed electronic structure including the

absence at high coverage of structure observed at low coverages and g) no

observed vibrational levels which correspond to free H2 molecule

Following is a short review of several proposed surface models

18Estrup and Anderson account for their LEED spot pattern results

using a model wherein the hydrogen molecules dissociate upon adsorption and

migrate initially to form islands of hydrogen atoms located at type A sites in

a e(2x2) structure Dissociation is assumed since the e(2x2) pattern maximizes

15 2at a coverage of 25x10 mo1ecu1escm whereas without dissociation it would

maX1m1ze at 5 1015 mo ecu1es cm2bull e site adsorptionx 1 Although type B or type

could also produce the initial e(2x2) LEED pattern these sites are rejected

15 2because of the following 1) There are only 1x10 B or e sitescm This is

not enough sites to accommodate a full monolayer coverage if all binding is

atomic and if it is into equivalent sites as suggested by the linear work

17

function change with coverage 2) While a complete 14-order LEED pattern is

possible with type B or type C sites it is not possible to produce a 14-order

54pattern with (012) spots missing using B or C sites As the adsorption

14 2proceeds past a surface density of 5xlO hydrogen atomscm the adatoms conshy

tinuously rearrange into long period superlattices with antiphase domains

that produce successively the 12-order spot splitting the 13-order streak

(only partly ordered domains) the 14-order pattern and finally the (lxl)

full coverage structure

This model has been criticized for not accounting for the two

observed binding states and for predicting a maximum in intensity of the half

order spots at e = 025 these beams maximize at less than that coverage In

addition the authors themselves questioned the idea of well-ordered super-

lattices

In order to explain the relative intensity of the two states at

saturation seen in flash desorption n~~ = 21 Tamm and Schmidt7

propose a 1 2

model wherein the initial hydrogen adsorption is atomic and at B sites in a

C(2x2) pattern At full coverage the remaining B sites are filled with

hydrogen molecules B sites are chosen over C sites using molecular orbital

55arguments A sites are rejected as there are too many to satisfy this model

This model satisfies the second order kinetics observed for the ~2 state and

the first order kinetics observed for the ~l state In addition it explains

14 2 7 their saturation coverage of gt 3xl0 moleculescm One weakness of this

model pointed out by Tamm and Schmidt is that it is not obvious the model can

account for the observed LEED and work function results Also no molecular

vibrations have been observed

18

19yonehara and Schmidt propose a three bonding state model that is

more consistent with the low temperature LEED results In their model an

additional state labeled ~3 is always present at the surface in A or B

2sites with a density of 5xl014 atomscm bull This state orders on the surface at

low temperatures to give the C(2x2) pattern observed At room and higher

temperatures it goes into solution (or perhaps disorders) becoming invisible

to LEED In the type A site model adsorption from the background fills in the

remaining A sites sequentially with the ~2 and ~l states This model is similar

18to that proposed by Estrup and Anderson except that the ~3 state although

invisible to LEED at room and higher temperatures occupies 14 of the type A

sites This model eliminates the stoichiometric objection to Estrup and

Andersons model The model depends on the possibility of type A sites being

occupied by states with considerably different properties If the ~3 state

occupies type B sites ~2 state adsorption fills in the remaining type B sites

and then ~l hydrogen adsorbs into random or possibly type C sites This

second model does not explain satisfactorily work function or LEED spot pattern

experimental data

56Propst found using a point defect type calculation that only a

B-site model of adsorption would produce the observed vibrational energies of

5769 meV and 135 meV Park using LEED intensity arguments found that only

adsorption into sites with 4-fold symmetry on the (100) surface (type B or type

C sites) would be consistent with broadening and splitting of the half-order

beams while the integral beams remain sharp

8Madey and Yates offer a critique of several of the proposed models